Performance of Weathering Steel in Bridges

Home , Tempering (metallurgy), Weathering steel

272

NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM REPORT 272 PERFORMANCE OF WEATHERING STEEL IN BRIDGES

TRANSPORTATION RESEARCH BOARD NATIONAL RESEARCH COUNCIL LJ ( i95 MAT. LAB. TRANSPORTATION RESEARCH BOARD EXECUTIVE COMMITTEE 1984

Officers

Chairman JOSEPH M. CLAPP, Senior Vice President, Roadway Express, Inc

Vice Chairman JOHN A. CLEMENTS, Commissioner, New Hampshire Department of Public Works and Highways

Secretary THOMAS B. DEEN, Executive Director, Transportation Research Board

Members RAY A. BARNHART, Federal Highway Administrator, U.S. Department of Transportation (ex officio) LAWRENCE D. DAHMS, Executive Director, Metropolitan Transportation Commission, Berkeley, California (ex officio, Past Chairman, 1983) MICHAEL J. FENELLO, Acting Federal Aviation Administrator, U.S. Department of Transportation (ex officio) FRANCIS B. FRANCOIS, Executive Director, American Association of State Highway and Transportation Officials (ex officio) WILLIAM J. HARRIS, JR., Vice President for Research and Test Department. Association of American Railroads (ex officio) DARRELL V MANNING, Director, Idaho Transportation Department (ex officio, Past Chairman 1982) RALPH STANLEY, Urban Mass Transportation Administrator, U.S. Department of Transportation (ex officio) DIANE STEED, National Highway Traffic Safety Administrator, U.S. Department of Transportation (ex officio) DUANE BERENTSON, Secretary, Washington State Department of Transportation JOHN R. BORCHERT, Regents Professor, Department of Geography, University of Minnesota LOWELL K. BRIDWELL, Secretary, Maryland Department of Transportation ERNEST E. DEAN, Executive Director, Dallas/Fort Worth Airport MORTIMER L. DOWNEY, Deputy Executive Director for Capital Programs, Metropolitan Transportation Authority, New York ALAN G. DUSTIN, President and Chief Executive Officer, Boston and Maine Corporation MARK G. GOODE, Engineer-Director, Texas State Department of Highways and Public Transportation LESTER A. HOEL, Hamilton Professor, Chairman, Department of Civil Engineering. University of Virginia LOWELL B. JACKSON, Secretary, Wisconsin Department of Transportation ALAN F. KIEPPER, General Manager, Metropolitan Transit Authority, Houston HAROLD C. KING, Commissioner, Virginia Department of Highways and Transportation FUJIO MATSUDA, President. University of Hawaii JAMES K. MITCHELL, Professor and Chairman, Department of Civil Engineering, University of California DANIEL T. MURPHY, County Executive, Oakland County Courthouse, Michigan ROLAND A. OUELLETFE, Director of Transportation Affairs, General Motors Corporation MILTON PII(ARSKY, Director of Transportation Research, Illinois Institute of Technology WALTER W. SIMPSON, Vice President-Engineering, Norfolk Southern Corporation JOHN E. STEINER, Vice President, Corporate Product Development, The Boeing Company LEO J. TROMBATORE, Director, California Department of Transportation RICHARD A. WARD, Director-Chief Engineer, Oklahoma Department of Transportation

TRANSPORTATION RESEARCH BOARD COMMITTEE FOR NRC OVERSIGHT (CNO) MILTON PIKARSKY, Illinois Institute of Technology (Chairman) JOHN R. BORCHERT, University of Minnesota JOSEPH M. CLAPP, Roadway Express, Inc. MARK G. GOODE, Texas State Dept. of Hwys. and Public Transp.

NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM

Transportation Research Board Executive Committee Subcommittee for NCHRP JOSEPH M. CLAPP, Roadway Express, Inc. (Chairman) RAY A. BARNHART, US. Dept. of Transp. JOHN A. CLEMENTS, New Hampshire Dept. of Public Works & Highways LAWRENCE D. DAHMS, Metropolitan Transp. Comm., Berkeley, Calif FRANCIS B. FRANCOIS, Amer. Assn. of State Hwy. & Transp. Officials THOMAS B. DEEN, Transportation Research Board

Field of Materials and Construction Area of Speccations, Procedures, and Practices Project Panel, D1O-22

CHARLES J. ARNOLD, Michigan Dept. of Transportation (Chairman) THEODORE H. KARASOPOULOS, Maine Dept. of Transportation HARRY BATSON, Federal Highway Administration ADO VALGE, Washington Metropolitan Area Transit Auth. BRUCE COSABOOM, New Jersey Dept. of Transportation JERAR NISHANIAN, FHWA Liaison Representative EARLE S. FREEDMAN, Maryland State Hwy. Administration WILLIAM G. GUNDERMAN, TRB Liaison Representative LOUIS A. GARRIDO, Louisiana Dept. of Transp. & Development

Program Staff

KRIEGER W. HENDERSON, JR., Director, Cooperative Research Programs ROBERT J. REILLY, Project Engineer LOUIS M. MACGREGOR, Administrative Engineer HARRY A. SMITH, Project Engineer CRAWFORD F. JENCKS, Projects Engineer ROBERT E. SPICHER, Projects Engineer R. IAN KINGHAM, Projects Engineer HELEN MACK, Editor NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM REPORT 272

PERFORMANCE OF WEATHERING STEEL IN BRIDGES

P. ALBRECHT and A. H. NAEEMI Sheladia Associates, Inc. ConsultIng EngIneers Riverdale, Maryland

RESEARCH SPONSORED BY THE AMERICAN ASSOCIATION OF STATE HIGHWAY AND TRANSPORTATION OFFICIALS IN COOPERATION WITH THE FEDERAL HIGHWAY ADMINISTRATION

AREAS OF INTEREST:

STRUCTURES DESIGN AND PERFORMANCE CONSTRUCTION GENERAL MATERIALS MAINTENANCE (HIGHWAY TRANSPORTATION) (PUBLIC TRANSIT) (RAIL TRANSPORTATION)

TRANSPORTATION RESEARCH BOARD NATIONAL RESEARCH COUNCIL WASHINGTON, D.C. JULY 1984 NATIONAL COOPERATIVE HIGHWAY RESEARCH NCHRP REPORT 272 PROGRAM Project 10-22 FY'82 Systematic, well-designed research provides the most effective ISSN 0077-5614 approach to the solution of many problems facing highway administrators and enginees. Often, highway problems are of ISBN 0-309-03851-0 lOcal interest and can best be studied by highway departments L. C. Catalog Card No. 84-51483 individually or in cooperation with their state universities and others. However, the accelerating growth of highway transpor- Price $12.00 tation develops increasingly complex problems of wide interest to highway authorities. These problems are best studied through NOTICE a coordinated program of cooperative research. The project that is the subject of this report was a part of the National Cooperative Highway Research Program conducted by the Transportation Research Board In recognition of these needs, the highway administrators of the with the approval of the Governing Board of the National Research Council. Such American Association of State Highway and Transportation approval reflects the Governing Board's judgment that the program concerned is of national importance and appropriate with respect to both the purposes and Officials initiated in 1962 an objective national highway research resources of the National Research Council. program employing modern scientific techniques. This program The members of the technical committee selected to monitor this project and to is supported on a continuing basis by funds from participating review this report were chosen for recognized scholarly competence and with due member states of the Association and it receives the full co- consideration for the balance of disciplines appropriate to the project. The opinions and conclusions expressed or implied are those of the research agency that per- operation and support of the Federal Highway Administration, formed the research, and, while they have been accepted as appropriate by the United States Department of Transportation. technical committee, they are not necessarily those of the Transportation Research Board, the National Research Council, the American Association of State Highway The Transportation Research Board of the National Research and Transportation officials, or the Federal Highway Administration, U.S. De- partment of Transportation. Council was requested by the Association to administer the research program because of the Board's recognized objectivity Each report is reviewed and accepted for publication by the technical committee according to procedures established and monitored by the Transportation Research and understanding of modern research practices. The Board is Board Executive Committee and the Governing Board of the National Research uniquely suited for this purpose as: it maintains an extensive Council. committee structure from which authorities on any highway The National Research Council was established by the National Academy of transportation subject may be drawn; it possesses avenues of Sciences in 1916 to associate the broad community of science and technology with the Academy's purposes of furthering knowledge and of advising the Federal communications and cooperation with federal, state, and local Government. The Council has become the principal operating agency of both the governmental agencies, universities, and industry; its relation- National Academy of Sciences and the National Academy of Engineering in the ship to the National Research Council is an insurance of ob- conduct of their services to the government, the public, and the scientific and engineering communities. It is administered jointly by both Academies and the jectivity; it maintains a full-time research correlation staff of Institute of Medicine. The National Academy of Engineering and the Institute of specialists in highway transportation matters to bring the find- Medicine were established in 1964 and 1970, respectively, under the charter of ings of research directly to those who are in a position to use the National Academy of Sciences. them. The Transportation Research Board evolved in 1974 from the Highway Research Board which was established in 1920. The TRB incorporates all former HRB activities and also performs additional functions under a broader scope involving The program is developed on the basis of research needs iden- all modes of transportation and the interactions of transportation with society. tified by chief administrators of the highway and transportation departments and by committees of AASHTO. Each year, spe- Special Notice cific areas of research needs to be included in the program are The Transportation Research Board, the National Research Council, the Federal proposed to the National Research Council and the Board by Highway Administration, the American Association of State Highway and Trans- the American Association of State Highway and Transportation portation Officials, and the individual states participating in the National Coop- Officials. Research projects to fulfill these needs are defined by erative Highway Research Program do not endorse products or manufacturers. Trade or manufacturers' names appear herein solely because they are considered the Board, and qualified research agencies are selected from essential to the object of this report. those that have submitted proposals. Administration and sur- veillance of research contracts are the responsibilities of the Published reports of the National Research Council and the Transportation Research NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM Board. are available from: The needs tor highway research are many, and the National Transportation Research Board Cooperative Highway Research Program can make significant National Research Council contributions to the solution of highway transportation problems 2101 Constitution Avenue, N.W. of mutual concern to many responsible groups. The program, Washington, D.C. 20418 however, is intended to complement rather than to substitute for or duplicate other highway research programs. Printed in the United States of America This report contains the findings of a comprehensive assessment of the perform- FOREWORD ance of weathering steel in bridges based on a review of the literature, a survey of By Staff practice in highway agencies and other organizations, and contact with selected in- Transportation dividuals knowledgeable on the subject. The report should be of value to structural Research Board engineers and others interested in the design, construction, and maintenance of steel bridges.

Weathering steel has been used in the construction of more than 2,000 bridges in the United States. Under the proper conditions, this material is expected to form its own protective oxide coating and to require no painting. It, therefore, offers the potential for considerable savings in life-cycle costs by elimination of the need for painting bridges, particularly bridges over major highways, electrified railways, or bodies of water. Under some environmental conditions, corrosion in weathering steel bridges has continued at a rate more rapid than anticipated, and there are concerns regarding the long-term performance of this material. Hence, the use of weathering steel in bridges has been more limited than might be expected in view of the potential cost savings. Information on the performance of weathering steel continues to accumulate, and there is a need to assemble and evaluate the data in order to facilitate the decision- making process and place it on a more rational basis. The objectives of NCHRP Project 10-22, "The Performance of Weathering Steel in Bridges," are (1) to assemble a systematic body of information on the performance of weathering steel, (2) to document and evaluate the current state of practice, and (3) to develop, where feasible, practical guidelines for design, construction maintenance, and rehabilitation of bridges using this material. The findings of the first phase of research are presented in this report and relate primarily to the first two objectives. In Phase I, problems that have developed in bridges were identified, and other factors that have limited the use of weathering steel in the United States and abroad were analyzed. The topics considered included: corrosion, salt exposure, other environmental factors, pitting, fatigue, in-service in- spectability, location within the bridge, structural details, shims and bearings, bolted connections, welding, adequacy of specifications, alloy content of the steel, initial cleaning and painting, remedial cleaning and painting, and adverse effects of corrosion on fatigue life. In addition, existing experimental and field performance data were assembled, and the current state of practice in regard to the use of weathering steel in bridges was evaluated. Some progress was made in Phase I on the third objective, but more research is needed. A second phase of research, initiated in August 1984, has as its specific objectives (1) to fatigue test 8-year weathered A588 transverse stiffener specimens under both constant and variable amplitude loading in both air and aqueous environments, and (2) to develop practical guidelines for design, construction, maintenance, and rehabilitation of weathering steel bridges. The findings contained in this report will be of interest to practicing engineers; however, the major recommendations based on both phases of research will be presented in the final report at the conclusion of Phase II. The information acquired in the first phase of research will serve, in part, as the basis for developing the Phase II guidelines, which are expected to be available in the form of specific recommendations to the AASHTO Subcommittee on Bridges and Structures by the spring of 1986. CONTENTS

SUMMARY

11 CHAPTER ONE Introduction Problem and Objectives...... 11 Historical Development...... 12 ASTM A242 Steels...... 14 ASTM A588 Steels...... 16 ASTM A709 Steels...... 18 Foreign Weathering Steels ...... 19

22 CHAPTER TWO Experiences Domestic Use ...... 22 DOmestic Experience ...... 22 European Experience ...... 27 Japanese Experience...... 28

30 CHAPTER THREE Corrosion Mechanisms Behavior of Steel in the Atmosphere ...... 30 Weathering Steel ...... 32 Effect of Aqueous Environments and Salt, 34 Effect of Primary Alloying Elements from Larrabee-Coburn Investigation...... 34 Effect of Alloying Elements from other Investigations.... 42 Prediction of Corrosion Penetration ...... 45

47 CHAPTER FOUR Atmospheric Corrosion Test Method ...... 47 Types of Environments ...... 48 Rural Environments...... 50 Industrial Environments ...... 53 Marine Environments ...... 58 Corrosion Resistance ...... 64

71 CHAPTER FIVE Service Corrosion Initial Climatic Condition...... 71 Shelter and Orientation...... 72 Angle of Exposure ...... 76 Continuously Moist Conditions ...... 78 Industrial Pollutants...... 80 Deicing Salts...... 83 Debris...... 85 Galvanic Corrosion...... 86 Pitting...... 88 Crevices ...... 88

90 CHAPTER SIX Strengthening Mechanisms and Toughness Introduction ...... 90 Strengthening Mechanisms...... 90 Alloying Elements...... 92 Toughness...... 93

97 CHAPTER SEVEN Weathering Fatigue Basis for Comparing S-N Data...... 97 Weathering Steels ...... 98 Nonatmospheric Corrosion Resistant Steels...... 108 Effect of Pitting .'...... 108 Comparison of Weathering Fatigue Data ...... 109

112 CHAPTER EIGHT Fatigue Design Corrosion Fatigue ...... 112 Estimation of Loss in Life ...... 116 Fatigue Design Stresses...... 119 Application to Current Designs ...... 123

125 CHAPTER NINE Painting Initial Painting of New Bridges ...... 125 Rust Stabilizing Treatment...... 128 Remedial Painting of Existing Bridges...... 128 131 CHAPTER TEN Connections Are Welding ...... 131 Specifications...... 132 Electroslag Welding ...... 132 High-Strength Bolted Joints ...... 134 Friction-Type Joints with Mill Scale' Surfaces...... 136 Friction-Type Joints with Blast-Cleaned Surfaces ...... 138 Pin and Hanger Connections ...... 141 Highway Guardrails...... 143 143 CHAPTER ELEVEN Structural Details Bridge Deck Joints ...... 143 Water and Debris Accumulation ...... 144 Staining of Abutments and Piers ...... 147 Recommendations from Ontario Study ...... 147 149 CHAPTER TWELVE Conclusion and Recommended Research Conclusions ...... 149 Recommended Research...... 149 150 REFERENCES

ACKNOWLEDGMENTS The research reported herein was performed under NCHRP Project President of Corrosion Consultants, Inc., supplied the historical back- 10-22 by Sheladia Associates, Inc.. Pedro Albrecht, Professor of Civil ground on the development of the weathering steels, checked the ac- Engineering, University of Maryland, College Park, Maryland, was the curacy of the exposure data, reviewed the discussions, and assisted in principal investigator. Amir H. Naeemi, Engineer, Sheladia Associates, some of the interpretations. Richard W. Vanderbeck, Engineering Ma- Inc., is the co-author of the report. terials and Processes, Inc., contributed to the material on strengthening Snehal Munshi, Senior Vice President of Sheladia Associates, Inc., is mechanisms and toughness. Kentaro Yamada, Associate Professor of the Coordinator and Project Manager on the project. Seymour K. Co- Civil Engineering, Nagoya University, Japan, provided the Japanese burn, Corrosion Engineer, retired, United States Steel Corp., and now experience and the bridge examples. PERFORMANCE OF WEATHERING STEEL IN BRIDGES

SUMMARY Weathering steel has been used nationwide in the construction of bridges. When the structural details are properly designed and the bridge is exposed in a suitable environment, weathering steel forms a protective oxide coating and should require no painting. Under these conditions, it offers the potential for considerable savings in life-cycle costs by the elimination of the need for maintenance painting. Under some environmental conditions cOrrosion has continued at a rate more rapid than anticipated, and there is concern about the long-term performance of this material. Hence, the expanded use of weathering steel in bridges has been limited despite the potential cost savings. The objectives of this research are to: (1) assemble a systematic body of information on the performance of weathering steel to date; (2) document and evaluate the current state of practice; and (3) recommend where feasible practical guidelines for design, construction, maintenance, and rehabilitation of weathering steel bridges. The findings presented in this report relate primarily to the first two objectives. Limited progress was made on the third objective, and a second phase of research, initiated in August 1984, has as its specific objectives (1) to fatigue test 8-year weathered A588 transverse stiffener specimens under constant and variable amplitude loading in air and aqueous environments, and (2) to develop practical guidelines for design, construction, maintenance, and rehabilitation of weathering steel bridges. The information acquired in the first phase of research will serve as the basis for developing the Phase II guidelines, which are expected to be in the form of specific recommendations to the AASHTO Subcommittee on Bridges and Structures.

History The weathering steels evolved in the 19 30's from a steel base that was alloyed with copper. The first major application came in 1933, in the form of exterior painted coal hopper cars, and later in the decade, in electrified railway and trolley cars. In 1948, a nonpainted weathering steel member was installed in a galvanized transmission tower. This first application was followed by the construction of unpainted transmission towers in several states. The first bridges and buildings were built from weathering steel in the mid- 1960's. The first bridges were: on Iowa 28 over Racoon River; on Eight-Mile Road over U.S. 10, Detroit, Michigan; over Brush Creek, Dayton Ohio; and in Morristown, as an adjunct to the New Jersey Turnpike. The first two bridges are now, 20 years later, scheduled for remedial painting. More information on the history of weathering steels can be found in Chapter One of the report, under "Historical Development."

Specifications The three weathering steel specifications ASTM A242, A588, and A709 were first issued in 1941, 1968, and 1974, respectively. The A242 Type 1 steels were mainl' used for architectural applications. The low-phosphorus version, A242 Type 2 steel, corresponds to the A588 and A709 Grade 50W steels. The A588 steels come in nine proprietary grades, each with a variation of the same basic chemical composition. A709 is a general specification for bridge steels. The two grades with enhanced atmospheric corrosion resistance, SOW and 100W, correspond to A588 steel and to a modified A5 14 alloy steel. The weathering steels come in yield strengths of 50 to 100 ksi (345 and 690 MPa). According to the ASTM specifications, they have an atmospheric corrosion resistance of approximately two times that of structural carbon steel with copper. This corresponds to four times that of structural carbon steel without copper (0.02 percent max. Cu). Matching high-strength bolts and weld rod are available in chemical compositions that give the connectors the same corrosion resistance and weathered appearance as those of the base metal. Few changes were made over the years in the weathering steel specifications. The most notable occurred in 1975, when the corrosion resistance of A242 steel was lowered from "at least" to "approximately" four times that of carbon structural steel without copper. This change aligned the corrosion resistance requirements for the A242 steel with those for the A588 and A709 steels (Chapter One, sections under "ASTM A588 Steels" and "ASTM A709 Steels"). The foreign weathering steel compositions have similar tensile properties and concentrations of those alloying elements that are added mainly to achieve corrosion resistance (Chapter One, "Foreign Weathering Steels").

Domestic Experience There are about 1,800 nonpainted weathering steel bridges on the federal-aid highway system, excluding state, county, city, tollroad, and mass transit bridges not on the federal-aid system. Weathering steel bridges were built in 43 states and the District of Columbia. Of those, 31 states and the District foresee using weathering steel for new bridges, 2 states are reconsidering their position and 10 states have stopped using it in a nonpainted condition. Many states carefully select the sites, avoiding severely polluted industrial environments, high humidity and rainfall areas, heavily salted highways and marine environments along the coast and waterways. For aesthetic reasons, Illinois, Missouri, and Wisconsin employ weathering steel mainly for stream crossings that are out of sight; Wisconsin uses it for grade separations in remote areas. Other than the requirement by some states to paint the steel structure up to 10 ft (3 m) on either side of deck joints, the design and detailing practice for nonpainted weathering steel bridges in the United States is the same as that for painted steel bridges. Of the 10 states that have stopped using weathering steel, Alabama, Indiana, Iowa, Michigan, Mississippi, and West Virginia have done so because contamination with highway deicing salts is accelerating the corrosion process. Alaska discontinued the use because of the cost of steel relative to concrete in that state, and because of the very humid environment along its southeastern coast. The long life of painting systems in the dry climate of Montana makes painted steel construction more economical than bare weathering steel construction. In North Dakota, concrete bridges are often more economical than steel bridges. Seven states never used bare weathering steel. Their main reasons are as follows. Arizona, Nevada, and New Mexico have dry climates in which painting systems last indefmitely, and weathering steels would not for a long time develop an oxide coating of uniform appearance. Steel is expensive in Hawaii, relative to concrete. New Jersey and South Dakota question the corrosion performance, particularly in salt environments, and dislike the rust staining of the concrete substructure. South Carolina has not used weathering steel because of problems reported by others. 3

Many steel users, including those who have rejected bare steel construction, have specified for economic reasons painted A588 steel either outright or as substitute for A572 in thicknesses greater than 2 in. (50 mm). This is done when the savings resulting from reduced weight, in comparison to A36 steel, exceeds the premium paid for high- strength low-alloy steel. The majority of bare weathering steel bridges built in the United States are performing in a satisfactory manner. But the application of weathering steel has not been without problems. In 1980, the Michigan Department of Transportation, following an extensive evaluation of their bridges, banned all uses of nonpainted A588 steel on the state highway system. They had found that traffic-sprayed runoff water contaminated with salt was severely corroding the bridges in rural and urban areas. In 1982, under the auspices of the AISI, a task group inspected 49 weathering steel bridges in Illinois, Maryland, Michigan, New York, North Carolina, and Wisconsin and on the New Jersey Turnpike. They found that 30 percent of the bridges showed good performance in all areas; 58 percent exhibited moderate corrosion in some areas; and 12 percent showed heavy corrosion in some areas. Most bridges, the report concluded, did not need immediate attention or overall painting. Notable exceptions are most of the structures inspected in the Detroit, Michigan, area. The corrosion problems arising from unsuitable applications of weathering steel are not limited to Michigan alone. There are known cases of bridges in Alaska, California, Iowa, Louisiana, Ohio, and Texas that have not performed to expectation. These bridges are not developing a protective oxide coating for two reasons: (1) salt contamination from any source; and (2) prolonged time of wetness. These bridges were not included in the AISI survey. Eight bridges in Ohio and one in Louisiana were remedially, painted in 1983. Many others in Alaska, California, Iowa, and Michigan are scheduled for cleaning and remedial painting. The domestic use of weathering steel and the experience with this material is described in Chapter Two, under "Domestic Use," and under, "Domestic Experience."

Foreign Experience The experience with weathering steel construction in Europe is reflected in the specifications such as those issued by France, Federal Republic of Germany, German Democratic Republic, and Czechoslovakia. All specifications caution against the deleterious effect that air pollution, .prolonged time of wetness, and salt contamination have on the corrosion performance. Several specifications describe the environments unsuitable for nonpainted weathering steel. In 1979, prior to Michigan's action, the West German Department of Transportation virtually banned the use of weathering steel for bridges on the federal system. Only two exceptions were granted; both were for composite girder bridges over electrified railways to avoid having to restrict the railway traffic during maintenance painting had the bridges been built of structural carbon steel. According to a survey conducted by the Japanese Society of Civil Engineers, 115 weathering steel bridges had been built in Japan by March 1979. Of those only 5 bridges were of nonpainted steel. The motive for using weathering steel prior to that date was mainly to prolong the life expectancy of the paint in industrial environments and coastal areas with relatively high humidity. The trend in Japan is shifting to bare steel construction, with some bridges receiving a sacrificial paint-like "weather coat" that helps the steel to form a protective oxide and prevents runoff water from staining the concrete substructure. Japanese steel companies have built and operated within their plants weathering steel bridges with painted, weather-coated and nonpainted sections. This allows them to closely monitor and compare the performance of the three systems. On the basis ru of the experience with these bridges, structural details for new bridges are being modified in ways that enhance drainage and inhibit accumulation of debris. These details are being implemented in new construction. The higher latitude and the generally higher levels of atmospheric pollutants cause the weathering steel to corrode more in the central and northern European countries than in the United States. The lower angle of the sun reduces the intensity of the drying cycle and prolongs the time of wetness. The Japanese use weathering steel despite having extensive marine exposures with high relative humidity and much rain. They try to Overcome the problem by careful design (Chapter Eleven, "Water and Accumulation") and selective use of a porous protective coating (Chapter Nine, "Rust Stabilizing Treatment"). A survey of weathering steel bridges in the Province of Ontario confirmed the observations of others, but gives more information on conditions existing within box beams. More information on the foreign experience can be found in Chapter Two, under "European Experience" and "Japanese Experience, and in Chapter Eleven, under "Recommendations from Ontario Study."

Corrosion Mechanisms The corrosion of metals proceeds by an electrochemical mechanism involving oxidation, reduction, an anode (corroding metal), a cathode (existing rust), and an electrical current through the electrolyte (water film). An oxidation reduction front separates the rust coating into an inner layer where rust is electrochemically reduced and an outer layer where magnetite is chemically oxidized to rust. Marked differences between the atmospheric corrosion behavior of structural carbon steel and weathering steel begin to show up only after a full rust coating has developed. Weathering steels derive their enhanced atmospheric corrosion resistance from the inclusion of 2 percent, or less, of such common alloying elements as copper, phosphorus, chromium, nickel, and silicon. On the basis of weight loss studies of steels with systematic variations of those elements, the following was found after 15.5 years of ideal and bold exposure in rural, industrial, and marine environments:

Increasing the copper content of a steel from 0.01 percent to 0.04 percent increased the corrosion resistance more than the addition of any other element. Further increases in copper were beneficial, but not to so great an extent as shown by the initial addition. Relatively small single additions of nickel, silicon, or phosphorus improved the corrosion resistance, but the largest improvements were obtained by specific combinations of these elements. The beneficial effects of several alloying elements were not additive.

Two mechanisms appear to explain the beneficial effects of these elements. One attributes the corrosion resistance to the precipitation of basic compounds of the alloying elements. The other has the alloying elements catalyze the precipitation of amorphous ferric oxyhydroxide. Both have several features in common: two-layer structure of the rust coating, electrochemical corrosion cycle, enrichment of alloying elements in the inner rust layer and basicity of the rust. Under prolonged periods of wetness and contamination with salt, a protective oxide coating does not form, and weathering steel corrodes at a rate comparable to that of structural carbon steel. Predictive equations fitted to extensive exposure data indicate that, even if the concentration of each element (Cu, P, Cr, Ni, and Si) in any grade of weathering steel were kept at the lowest level specified by ASTM, which is unlikely, these boldly exposed steels should still exhibit a 15.5-year corrosion penetration of not more than 6.4 mils (163 ,Lm) in the rural, 7.1 mils (180 m) in the industrial and 9.3 mils (236 m) in the marine environments tested. The penetrations reported for some severely corroding weathering steel bridges exceed these values much earlier, some after one year. This degree of corrosion cannot be explained in terms of a lean composition. It is caused primarily by a misapplication of the steel in environments different from the desired bold exposure. The corrosion mechanisms are presented in detail in Chapter Three.

Atmospheric Corrosion The corrosion resistance of a steel is typically determined from multiyear exposures of standard size coupons mounted on racks facing south at a 30-deg angle from the horizontal. The average corrosion penetration per side is calculated from the weight loss of specimen after the oxide layer is chemically removed. The macroenvironments at the exposure sites are broadly classified as being rural, industrial, and marine. Urban environments are often included with industrial environments because both have similar atmospheric pollutants. The standardized exposure conditions yield comparable data for macroenvironments, but they are not typical of bridge service conditions (Chapter Four, Test Method" and "Types of Environments"). Rural environments are usually free of aggressive pollutants. They contain relatively small amounts of sulfur oxides and carbon dioxide from combustion products. In such clean air, the rate at which steels corrode increases only when the relative humidity exceeds about 70 percent. Rural environments generally are not aggressive toward steel (Chapter Four, "Rural Environments"). The most potent causes of corrosion in industrial environments are the sulfur oxides and nitrogen oxides produced by burning automotive fuels and fossil fuels in industrial and utility power stations. The sulfur oxides dissolve in films of condensed moisture on the surface of the steel where they can be further oxidized to form sulfuric acid and attack the steel. This occurs at a critical relative humidity as low as about 60 percent. In the absence of moisture and condensing conditions sulfur oxides are not necessarily aggressive toward steel (Chapter Four, "Industrial Environments"). In marine environments airborne salt spray droplets, salt fog, and nightly condensate drops can keep the steel damp for long periods of time. On drying, salt crystals hydroscopically attract moisture at low relative humidities. Visible condensate, in the form of droplets, is not necessary for corrosion to proceed. The presence of chloride inhibits the role of sulfur oxides in forming a protective oxide film. Weathering steel may still resist corrosion provided that the surfaces are boldly exposed and rain frequently washes off the salt. However, the corrosion resistance is greatly diminished in severe marine environments, or when the surfaces are sheltered. Information gained from the performance of the weathering steels in marine exposure tests aid in determining the corrosion behavior of bridges contaminated with roadway deicing salt (Chapter Four, "Marine Environments"). While rural atmospheres are less corrosive than industrial atmospheres and much less than marine atmospheres, this broad classification by macroenvironment is not necessarily a reliable description of the aggressiveness of an atmosphere towards steel. The corrosiveness of an exposure site depends mainly on: (1) the time of wetness of the specimens, determined largely by the length of time during which the relative humidity exceeds a critical value; (2) degree of airborne sulfur oxide pollution; and (3) chloride contamination. The severity of these factors are greatly influenced by differences in the microenvironments of the exposure sites even though the macroenvironment may be the same. 6

For weathering steels exposed to rural and industrial environments, the curves of corrosion penetration versus time rise at first and then tend to level off. The curves for American sites fall within the scatterband of 2 to 4 mils (50 to 100 m) corrosion penetration reported for Mayan R steels (Bethlehem Steel) with a broad range of chemical compositions, after 18 years exposure in different environments. The penetrations at some of the generally more corrosive German and English sites exceed 4 mils (100 sm), some after only 2 years (Chapter Four, under "Rural Environments" and "Industrial Environments"). For weathering steels exposed to east and gulf coast environments, the corrosion penetration curves continue to rise at a constant rate. Most exceed 4-mils (100-.tm) penetration after 5 to 10 years exposure. Those for European sites rise even faster. The constant presence of salts in marine environments inhibits the formation of a protective oxide coating (Chapter Four, "Marine Environments"). When using bold atmospheric exposure data to estimate the corrosion performance of bridges built from weathering steel, the effects of microenvironment associated with local topography, climate, and service conditions must be carefully considered, along with the fact that bold exposure does not occur under a bridge deck.

Corrosion Resistance - The ASTM specifications A242, A588, and A709 state that the atmospheric corrosion resistance of weathering steels is about equal to two times that of carbon structural steel with copper, which is equivalent to four times carbon structural steel without copper (Cu 0.02 percent max.). But they do not defme how the corrosion resistance of the weathering steel, relative to the reference steel, should be determined. Ratios of corrosion penetrations, rates, and times have been used for this purpose, depending on the viewpoint and application. The ASTM results do not identify or define good and poor corrosion performance. Furthermore, the advantages of the more corrosion resistant weathering steels tend to be understated in the less corrosive environments. The observed anomalies come from tying the behavior of weathering steel to that of carbon and copper steels (Chapter Four, "Corrosion Resistance"). Consideration should be given to adopting design guidelines that state the requirements for the corrosion resistance of weathering steels in terms of the penetration, independent of the behavior of the reference steels. They should specify: (1) curves of allowable corrosion penetration per side versus exposure time in different environments; (2) conditions under which material thickness should be increased to compensate for net section loss; and (3) required increase in thickness. Some foreign specifications have taken this approach.

Service Corrosion The microenvironment and the service conditions encountered at bridge sites can alter the corrosion behavior of boldly exposed weathering steel test specimens. The main factors are: initial climate, sheltering, orientation, angle of exposure, time of wetness, atmospheric pollutants, deicing salt, and debris. Depending on the service conditions and type of details, the steel may be subjected to the following forms of corrosion not previously discussed: poultice, galvanic, pitting, and crevice attack. Initial Climatic Conditions. The climatic conditions at initial exposure can produce large differences in corrosion penetration. These differences are much larger for carbon steel than for weathering steel. For neither type of steel, however, does the difference disappear with increasing exposure time. Evidently, the initial reactivity of the steel with its environment affects the long-term corrosion behavior (Chapter Five, "Initial 7

Climatic Condition"). For these reasons, initially contaminated weathering steel should be cleaned after installation. This was done in the case of the Luling Bridge, Louisiana, many of whose sections had been exposed to sea-salt spray during transport from Japan. Orientation. Specimen surfaces that are exposed to the sun and rain corrode less. Those facing the ground, the north (in the northern hemisphere), and sources. of contaminants corrode more (Chapter Five, "Shelter and Orientation"). Sheltering. The corrosion penetration of test specimens exposed under a bridge was found to be greater than that of control specimens boldly exposed. Similarly, because of better drainage and less debris accumulation, sheltered vertical specimens performed better than sheltered horizontal specimens (Chapter Five, "Shelter and Orientation"). Under conditions of bold exposure, vertical and horizontal specimens corroded more than specimens facing south at an angle close to the latitude of the site, a condition that maximizes sun drying and rain washing (Chapter Five, "Angle of Exposure"). Continuously Moist Condition. Alternate cycles of wetting and drying are essential to the formation of a protective oxide coating. When the weathering steels remain continuously wet, they do not exhibit corrosion resistance superior to that of carbon steel. Direct precipitation of rain, snow, or condensation is not needed for a steel surface to remain continuously wet. Moisture can be deposited by capillary action of the porous oxide coating, and absorption by corrosion products or salt deposits on the surface. A thin, invisible film of electrolyte can form on the surface at a relative humidity well below 100 percent. The critical relative humidity, above which the corrosion rate increases sharply, depends on the nature of the atmospheric pollutants, surface contaminants, and corrosion products. Weathering steels with a dense oxide coating in a salt-free environment corrode beginning at relative humidities somewhat higher than 65 percent, while those with a loose oxide coating in a salt-laden environment can corrode down to 55 percent relative humidity. The length of time during which the critical relative humidity is exceeded determines the time of wetness and greatly affects the corrosion penetration. Reducing the degree of contamination and the time of wetness is a requisite for improving corrosion performance (Chapter Five, "Continuously Moist Conditions"). Industrial Pollutants. The proximity to a source of industrial pollutants can alter the corrosion behavior of weathering steels. This depends on the type and concentration of atmospheric constituents emitted by the source. The sulfur oxides are the most widespread and potent cause of corrosion in industrial environments. However, with the advent of the EPA standards, the installation of scrubbers in fossil fuel plants and the use of low-sulfur coal, excessive levels of sulfur oxides are less likely to occur now and in the future. Among the local sources that emit pollutants are the chioralkali plants, chemical plants, pulp and paper plants, fertilizer plants, oil refmeries and petrochemical plants as well as coal-fired electrical generating facilities. To assess the corrosiveness of a localized area towards the weathering steels, specimens of carbon steel and weathering steel should be exposed for a minimum period of 18 to 24 months. Either the 30 deg to the horizontal or the vertical mode of exposure may be used. Deicing Salts. Deicing salts spread on highways and bridge decks can create an aggressive localized environment that often exceeds conditions prevailing near the ocean. The corrosive deicing salt solution can leak through the bridge deck by way of expansion and construction joints and drain along girder and floor beam flanges leaving a corrosive salt residue that remains active throughout the year by absorbing moisture from the air. In the case of double deck bridges and overpasses, a salt mist 8

or spray from moving traffic can be generated such that it splashes overhead girder structures leaving a salt residue that remains active throughout the year. Likewise, dried salt residue from the highway can be blown into and on various bridge components where it can continue its aggressive action during humid periods. Salt water also moves upgrade on flanges and around "drip stops" or intended run-off points by, that same capillary action. Evaporation of the water leaves salt deposits, leading to accelerated pitting of the surface. Such structures should be high-pressure hosed or stream-detergent cleaned at the close of each winter season. Areas experiencing aggressive attack will require spot protection on a periodic basis (Chapter Five, "Deicing Salts"). Poultice Corrosion. The debris that accumulates on the horizontal surfaces of girder elements consists of road dirt, salt, leaves, rust flakes spalling from the web, and occasionally bird nests and droppings. By retaining moisture, the debris forms a poultice that keeps chlorides and sulfates in close contact with the steel and greatly prolongs the time of wetness (Chapter Five, "Debris"). Galvanic Corrosion. When two dissimilar metals are in contact in a moist environment, a galvanic couple forms and causes the anodic metal to corrode more than the cathodic metal. Steel is anodic to mill scale. Thus, mill scale can cause galvanic pitting of the steel along breaks in the mill scale. Carbon steel is anodic to low-alloy steel, and structural steels are anodic to more noble metals such as bronze (Chapter Five, "Galvanic Corrosion"). To avoid galvanic corrosion in aggressive environments, bolts and weld filler metal should have compositions similar to the weathering steel. Bronze washers in pin-and-hanger connections must not be in direct contact with the steel. Pitting. In all instances the chloride ion from deicing salt or seawater spray are the primary stimulants that accelerate pitting. Differential aeration cells resulting from the scattered deposits of debris, likewise, stimulate pitting of the surface of structural steels. Under these conditions, bridges covered with mill scale may develop numerous pits at cracks in the scale. Bridges that are free of mill scale can, however, develop deep pits beneath deposits of moist debris (Chapter Five, "Pitting"). Crevices. Crevice corrosion is a form of localized attack of the steel surface at, or immediately adjacent to, an area shielded from bold exposure. Bolted connections, cantilever expansion joints, guardrail lap joints, and lap joints not welded all around are examples of details susceptible to crevice corrosion. The rate of corrosion attack in the crevice is higher than that on the surfaces outside the crevice (Chapter Five, "Crevices").

Fatigue The available test data on fatigue of weathering steel structural details come from specimens that were exposed to the environment for a number of years and then fatigue-cycled to failure in dry laboratory air. This condition, termed herein weathering fatigue (Chapter Seven), reduced the life of most details and increased the life of a few. The observed reduction of life, attributed to rust pitting, was greatest for base metal and continuously decreased as the stress concentration in the detail increased. The observed gain in life of a few details resulted from weld-toe rounding by corrosion. The aforementioned behaviors were the same for weathering steels and carbon steel. It should be kept in mind that the latter would be coated and not subjected to pitting. In addition to the effect of weathering, bare steel bridges are subjected to corrosion fatigue during their service life (Chapter Eight). All available corrosion fatigue and crack growth rate data show that cracks initiate earlier and propagate faster in aqueous environments than they do in air. There are no data that show a beneficial effect of 9

exposure to aqueous environments. Therefore, it may be assumed that weathering steel bridges have shorter fatigue lives than painted bridges. The authors' estimates, accounting for the effects of both weathering and corrosion fatigue, give the following total losses in stress range: 34 percent for Category A base metal, 24 percent for Category B welded beams, 19 percent for Category C transverse stiffeners, 13 percent for Category C 2-in. (51-mm) attachments, and 11 percent for Category D, E, and E' attachments. The corresponding losses in terms of number of cycles are: 73 percent for Category A base metal, 60 percent for Category B welded beams, 47 percent for Category C transverse stiffeners, 37 percent for Category C 2- in. (51-mm) attachments, and 29 percent for Category D, E, and E' attachments (Chapter Eight, under "Estimation of Loss in Life" and "Fatigue Design Stresses"). The AASHTO fatigue specifications are based on fatigue test data obtained in clean and dry laboratory air. Current design practice applies these provisions to both painted steel and weathering steel bridges. The new drafts of the European codes for fatigue design of steel structures state that the "reduction in life due to corrosion" is not explicitly covered. Their design S-N curves are said to apply only to structures in normal atmospheric , conditions with suitable corrosion protection. British fatigue specifications for offshore structures impose a penalty factor of two on the fatigue life (50 percent loss in life) for all steel welded joints exposed to salt water unless they are coated or cathodically protected (Chapter Eight, "Fatigue Design Stresses"). The environment of a bridge is different from offshore structures but, in some ways, not less severe because chlorides build up on the surfaces of salt-contaminated members that are sheltered and, therefore, not rain washed. It is prudent to design bare weathering steel structures for fatigue to reduced allowable stress ranges. Additional fatigue tests of weathered steel members are currently being sponsored by the Maryland Department of Transportation. The second phase of research under NCHRP Project 10-22 also includes tests of 8-year weathered weldments. The results of these studies will be considered in developing fatigue design recommendations in the second phase of research.

Painting Initial paint systems last longer on a weathering steel substrate, because the protective and dense oxide coating that forms in scratches and exposed areas does not undercut the adjacent intact paint film. With carbon steel, the expanding rust continuously destroys the adhesive bond between the paint and the steel surface. The procedures and standards for preparing the surface of new weathering steel are the same as those for carbon steel. With chlorides present, paints do not perform better on A588 steelthan on carbon steels (Chapter Nine, "Initial Painting of New Bridges"). In comparison with shop coating new steels, the problem with the remedial painting of severely corroded and pitted weathering steels lies in removing all rust and salt trapped in dimples and pits. Unless this is achieved, the paint life is reduced. With proper surface preparation and paint application it is possible to protect A588 steel against corrosion, even if it has been exposed to a severely leaking bridge joint environment. Zinc-rich primer compositions appear to out perform other systems (Chapter Nine, "Remedial Painting of Existing Bridges"). Final recommendations for remedial painting of weathering steel bridges must await the results of on-going studies being performed by the Michigan Department of Transportation and the Steel Struc- tures Painting Council.

Connections Weathering steels are weldable with the use of good welding practice and the proper choice of filler metal as specified by the American Welding Society. Covered carbon 10

steel electrodes and bare carbon steel electrode-flux combinations may be used for: underlying passes of multiple-pass weidments, single-pass shielded metal arc welds to 0.25-in. (6.4-mm) size, single-pass submerged arc welds 5/16-in. (8-mm) size, and painted structures. Covered low-alloy steel electrodes and bare low-alloy steel electrode-flux combinations are used for bare applications requiring weld metal with atmospheric corrosion resistance and coloring characteristics similar to the base metal (Chapter Ten, under "Arc Welding" and "Specifications"). A325 and A490 Type 3 high-strength bolts have atmospheric corrosion resistance and weathering characteristics comparable to that of weathering steel. As is done in welds, matching the chemistry of the connectors and the base metal prevents accelerated galvanic corrosion of the connectors. The crevice between the plies of a bolted weathering steel joint usually seals itself with oxide when the joint is appropriately tightened and stiff. If the joint moves, the surfaces should be coated with a protective material and filled with a suitable sealant to avoid progressive corrosion. Expansion forces induced by corrosion products in the crevices can deform the connected elements and cause large tensile loads on the bolts. Significant bowing of the plates between bolts or edge distortion can be controlled by limiting the pitch to 14 times the plate thickness or 7 in. (178 mm) and the edge distance to 8 times the plate thickness or 5 in. (127 mm), whichever is smaller (Chapter Ten, "High-Strength Bolted Joints"). Friction-type connections of weathering steel plates with mill scaled surfaces have much lower slip resistance than their carbon steel counterparts. The loss is attributed to the dense, hard and slippery nature of the mill scale on HSLA steels. Corrosion of the mill scale contact surfaces before the joint is high-strength bolted tends to increase the slip resistance (Chapter Ten, "Friction-Type Joints with Mill Scale Sur- faces"). Friction-type connections with blast-cleaned surfaces have a slip resistance comparable to that of blast-cleaned carbon steel plates (Chapter Ten, "Friction-Type Joints with Blast-Cleaned Surfaces"). Loose and flexible joints are prone to crevice corrosion and packout bulging. Such joints must be protected against corrosion and provisions made for cleaning and maintaining them. Examples of such joints known to exhibit the aforementioned problems are pin-and-hanger connections and guardrail lap joints (Chapter Ten, under "Pin and Hanger Connections" and "Highway Guardrails").

Structural Details The most important considerations when detailing weathering steel bridges are to divert the flow of runoff water from the structure, prevent ponding, and avoid the accumulation of debris that traps moisture containing chlorides. Unfortunately, leaking of expansion joints has been one of the major causes of extensive maintenance and costly remedial work on bridges. Experience has demonstrated that the performance of many joint systems is disappointing. To prevent the progressive corrosion of weathering steel bridges, several states have been painting the girder ends on either side of an expansion joint over a length varying from one beam depth to 10 ft (3.0 m). Water ponds and debris accumulate on horizontal surfaces and in corners formed by horizontal and vertical plates. The most susceptible locations are: the top surfaces of lower beam flanges and truss chords, gusset plates for horizontal bracing, bolted splices of horizontal and sloped members, bearing and intermediate stiffeners, and the interior of nonsealed box sections. To avoid water ponding and debris accumulation it is important to: (1) slope horizontal surfaces longitudinally or transversely; (2) avoid re-entrant corners; (3) seal box sections; and (4) design details for easy discharge of water. To enhance the good performance of weathering steel bridges, careful attention 11

must be given to detailing. It is important to note that salty water causes the steel to develop a porous oxide coating with high capillarity. As a result, water migrates to places it would not normally reach. This may reduce the effectiveness of some details designed to remove water from the surface. Chapter Eleven gives more information on structural details for weathering steel bridges and their implementation in two bridges.

Conclusion The majority of the weathering steel bridges are in good condition, although localized areas of accelerated attack can be found in many structures. A number of bridges in several states are experiencing excessive corrosion because of salt contamination and/ or prolonged time of wetness. Weathering steel can provide a satisfactory service life with limited maintenance if the structural details are designed in a manner that prevents accelerated attack, vulnerable areas are painted, and contamination with chlorides is avoided. The report contains extensive information on the performance of weathering steel and its use in highway bridges. The practitioner can use this body of knowledge as a general guide for designing new bridges and for trouble shooting. The second phase of research under NCHRP Project 10-22 will develop specific guidelines for design, construction, maintenance, and rehabilitation of weathering steel bridges.

CHAPTER ONE

INTRODUCTION

PROBLEM AND OBJECTIVES Despite the aforementioned advantages, weathering, steel is recently being specified for new bridges less than might be Problem expected. The main reason for this stems from the observation and the concern that some bridges are corroding faster than Weathering steel specimens boldly exposed under ideal con- anticipated. This may result in a significant loss of metal not ditions develop a dense and well-bonded oxide coating that considered in design. shields the underlying steel base against penetration of moisture, Weathering steels develop a protective oxide coating under oxygen, and atmospheric contaminants, and greatly reduces the following conditions: long-term corrosion. When properly designed and exposed in a suitable environment, weathering steel bridges are also expected Atmospheric exposure to intermittent wet-dry cycles with- to for-in a protective oxide coating and to require no painting. out prolonged wet periods. Weathering steel offers the potential for savings in life-cycle Absence of heavy concentrations of corrosive pollutants, costs. The initial cost of an unpainted weathering steel bridge especially deicing salts. is said to be about the same as the cost of a painted regular Washing of the exposed surface by rain water. steel bridge. Because a weathering steel bridge should not need Absence of detail geometrics that trap moisture, dirt, or maintenance painting, its life-cycle cost would be lower than debris and hence foster corrosion. that of a painted steel bridge. Furthermore, by eliminating the need for maintenance painting, paint jobs of bridges over difficult These conditions may not be met in some bridges. For ex- terrains, electrified railways, and bodies of water can be avoided; ample, the deck shelters the steel girders from the rain. Runoff heavy traffic lanes need not be closed; and safety hazards to water, mixed with roadway deicing salts, leaks through joints motorists and painters alike are prevented. In an evaluation of onto the steel. Salt-bearing sprays kicked up by trucks passing 16 factors that optimize the cost of steel plate girders for con- under a bridge settle on the girders. Roadway debris and rust tinuous composite bridges with span lengths up to 200 ft (60 flaking off the web accumulate on the flange and hold moisture. m), Knight [1983] cited the use of unpainted weathering steel Water collects at poorly designed structural details. A lack of as being one of the three most influential factors. aeration and low clearance over bodies of water keep the steel 12 damp over long periods of time. When some of these conditions As a consequence of numerous reports of copper additions occur at a bridge site, corrosion problems can develop. to iron during the 19th century, Williams conducted a simple In addition to the previously described potential corrosion experiment in 1900. He selected a series of copper-bearing and problems, there have been concerns about localized galvanic and copper-free wrought irons and Bessemer steels and wetted them crevice corrosion, reductions in life due to rust pitting and with water several times daily for 1 month while exposed out- corrosion fatigue, and rust staining of concrete piers and abut- doors. Only the copper-bearing products significantly retarded ments by oxide-laden water running off the steel. corrosion. The subject of copper in steel and iron was frequently dis- Objectives cussed during the first decade of the existence of the newly formed American Society for Testing and Materials (1900- The objectives of this study are to: (1) assemble a systematic 1910). It remained forG.H. Clamor at the 1909 meeting to call body of information on the performance of weathering steel; (2) attention to the work of Burgess and Aston (also members of document and evaluate the current state of practice; and (3) the Society) in which they added single elements to pure elec- develop, where feasible, guidelines for design, construction, trolytic iron and exposed it in Madison, Wisconsin. They found maintenance, and rehabilitation of weathering steel bridges. that copper enhanced the corrosion performance of the iron. Bridges are subjected to a variety of service conditions and Clamor further reported that copper improved the ultimate many complex factors which influence the performance of strength of carbon steel. He suggested that Sudbury (Ontario, weathering steel. Recognizing that all bridges cannot be designed Canada) ores contained nickel and copper, and that such natural for good performance with a general set of guidelines, the report ores should provide a steel with improved properties without presents data from which the designer can extract the infor- any further alloy additions. mation needed for assessing the performance under a given set The most authoritative review of the state of the art is con- of conditions. The information available in the open literature tained in a report in the 1915 Yearbook of the American Iron is adequate to evaluate the performance of weathering steel. and Steel Institute by D.M. Buck of the American Sheet and The findings presented in this report relate primarily to the Tin Plate Company, a division of the United States Steel Cor- first two objectives. Limited progress was made on the third poration. He reported that he observed in 1910 the excellent objective, and a second phase of research, initiated in August performance exhibited by a 0.07 percent copper-bearing steel 1984, has as its specific objectives (1) to fatigue test 8-year sheet. To confirm that observation, full-size 16-gage and 27- weathered A588 transverse stiffener specimensI under constant gage roofing sheets of carbon steel with varying copper levels and variable amplitude loading in air and aqueous environments, were exposed in rural, industrial, and marine atmospheres. and (2) to develop practical guidelines for design, construction, Buck reported his findings in 1913 in a paper given before maintenance, and rehabilitation of weathering steel bridges. the American Chemical Society in which he stated that no The information acquired in the first phase of research will difference exists between sheets with 0.15 percent and 0.34 per- serve as the basis for developing the Phase II guidelines, which cent copper content; whereas, sheets without copper rusted are expected to be in the form of specific recommendations to through during the first year of exposure in a coke oven at- the AASHTO Subcommittee on Bridges and Structures. mosphere. The sheets with 0.06 percent copper took an inter- The remainder of this chapter traces the history of the de- mediate position. velopment of weathering steels and reviews the applicable Buck, with the assistance of Handy, then director of research ASTM specifications. Chapter Two summarizes the domestic for the Pittsburgh Testing Laboratories, performed a more de- and foreign experiences. Chapters Three to Five deal with the tailed exposure test in which they exposed around Pittsburgh corrosion performance, and Chapter Six with strength and im- full-size roofmg sheets that varied in copper levels from 0.04 pact properties. The fatigue behavior is examined in Chapters percent to 2.2 percent. In addition, they obtained a number of Seven and Eight. This is followed by Chapter Nine on initial "pure" irons which contained from 0.014 percent to 0.018 per- and remedial painting, Chapter Ten on welded, bolted, and cent copper. At the end of almost 1 year they found that the miscellaneous connections; and Chapter Eleven on structural low copper level specimens lost weight at the rate of 0.31 oz/ details. Chapter Twelve includes the conclusions and an enu- ft2/mo (95 g/m2/mo) in contrast to a loss of 0.12 oz/fe/mo meration of areas needing further investigation. (37 g/m2/mo) for all of the copper-bearing specimens. Thus they were able to locate the point for effective copper levels in carbon steel. HISTORICAL DEVELOPMENT Allerton Cushman of the Institute of Industrial Research in Washington, D.C., and a member of ASTM Committee A5, This brief history traces the development in the United States voiced objection to the facts reported by Buck and Handy at of what are familiarly known as the "weathering steels" of which the American Iron and Steel Institute meeting in 1915. He the most promising application currently is in highway bridges. insisted that a carefully supervised test be repeated under the They were part of a family of steels termed "high-strength low- sponsorship of Committee AS of the ASTM with cooperation alloy" (HSLA). The products evolved from an iron and steel from the National Bureau of Standards. Furthermore, the ex- base. It was noticed in the early 1800's, and later reported by posure tests were to be located at protected sites such as Fort Karston in 1827, in Germany, that additions of small amounts Sheridan, Illinois; the Naval Academy at Annapolis, Maryland; of copper to foundry products resulted in a reduction in the and at Fort Pitt in Pittsburgh, Pennsylvania. corrosion rate in acid. Because of this observation, many people During the preparation of the test specimens, it was pointed were trying to evaluate and predict the performance of different out (in the ASTM Committee AS report in 1916) that the iron steel compositions for use in structural applications by the rate specimens have a high copper content because of the use of at which they dissolved in 20 percent sulfuric acid. eastern Pennsylvania ores and that an effort would be made to 13 find ores substantially free of copper such as typified by the rosion resistance and strength, among many possible combi- newly available Mesabi Range ores. nations, has resulted in a steel four to eight times more corrosion The Proceedings of the American Society for Testing Mate- resistant than carbon steel in the bare state in a variety of rials for 1919 contain Buck's report in which steels with copper aggressive environments. The steel has a minimum yield strength levels varying from 0.01 percent up to 0.25 percent were exposed of 50 ksi (345 MPa) in 1/2-in. (12-mm) thick plates for com- at the aforementioned test sites and inspected twice yearly. Com- position containing about 0.10 percent carbon. Similar yield mittee members included Buck and Cushman or his represent- strengths were also achieved through 4-in. (100-mm) thick plates ative. The fmdings following 18 months of exposure indicated by increasing the carbon and manganese levels and adding va- that beginning with 0.04 percent copper the atmospheric cor- nadium. rosion rate slowed down considerably, and that a copper level USS Cor-Ten steel, the first and best known of the weathering of 0.15 percent was ample. type high-strength low-alloy steels were first marketed in 1933. Its acronym, Cor-Ten, was derived from the two properties that The results of this 1916 exposure test were summarized and presented again by Buck at a meeting of the American Iron and distinguish it from mild carbon steel and copper-bearing steel; namely, improved atmospheric corrosion resistance and higher Steel Institute and published in their Yearbook for 1920. In that article Buck referred to the earliest use he could fmd for copper, yield and tensile strengths. namely, in 1627, wherein the physical properties of the iron The first major application in 1933 was in the form of an exterior painted coal hopper car. This met with instant success, were improved. He related that Richardson and Richardson, in and since then over 1,000,000 such cars have been fabricated 1916, reported that copper additions to steel resulted in greater and placed in service worldwide. The next major application at improvement than equal copper additions to pure iron. that time came as a result of the Great Depression. It resulted On the strength of their own atmospheric exposure tests, the in the standard design for an electrified railway or trolley car New York Central Railroad purchased 620,000 copper-bearing known as the PCC car. This stands for the President's Confer- steel tie plates weighing 4,000 tons and containing 0.2 percent ence Car, a result of President Roosevelt's effort to improve the copper based on their findings of a loss rate one-tenth that of nation's transit system while providing an economic stimulus plain carbon steel plates. Buck states unequivocally in his report to the steel industry. Many such cars went into service in the that the superiority of copper-bearing steels is twice that of plain 1936-1939 period with the remainder supplied in the 1946- carbon steel and, in fact, his findings show it to be sometimes 1949 period. The city of Pittsburgh, Pennsylvania, is currently as much as three to five times better in certain aggressive en- rehabilitating 45 such trolleys from the 1949 class. With the vironments. exception of dents, the cars continue in excellent condition, and The U.S. Steel Corporation, through its American Sheet and are retaining the steel between the belt line and the roof. Tin Plate Division, began marketing copper-bearing steel sheets The major patents covering Cor-Ten steel expired in 1940 in 1911. At about this time, Byramji D. Saklatwalla came from and 1950, respectively. However, the trademark was retained England to Pittsburgh, Pennsylvania (1909) to investigate means and constitutes one of the major marketing tools for United for incorporating vanadium into carbon steel. He was not un- States Steel Corporation. Just about all of the major steel firms aware of the activities of Buck and began some experiments in developed a variation of the Cor-Ten steel composition for ap- this field. By 1926, in cooperation with one of the U.S. Steel plications where atmospheric corrosion resistance was desired. mills (Vandergrift, Pennsylvania), he obtained a patent covering The first ASTM specification entry was in 1941 with A242-41T the addition to carbon steels of several elements besides copper for Low-Alloy Structural Steel. It covered steel suitable for to both enhance atmospheric corrosion resistance as well as to welding or riveting intended primarily for use as main stress- increase the yield strength beyond that achieved by copper; carrying structural members. The scope was expanded in the namely, 32 to 36 ksi (220 to 250 MPa). Meanwhile, during the 1942 edition to cover applications where savings in weight and period 1920-1930 U.S. Steel was developing a family of high- atmospheric corrosion resistance were important. The degree of strength low-alloy steels intended primarily for the railroad in- atmospheric corrosion resistance provided by A242 was quan- dustry. tified in the 1960's as being at least two times that of structural At the same time in Pittsburgh, Jerome Straus, paralleling carbon steel with copper and four times that of structural carbon Saklatwalla's work, also developed a high-strength low-alloy steel without copper (0.02 percent copper content). steel composition for which a patent was granted in 1935. U.S. Following the commercialization of the HSLA steels during Steel acquired the patent rights of both of these inventors to the 1930's, a number of research reports and reviews described round out their product line of HSLA steels. the superior atmospheric performance that can be expected from An article by Cone in Steel Magazine, in 1934, reviewed the Cor-Ten steel over copper-bearing steel [Schranim 1934a; dozen or so new proprietary HSLA steels then available and Schramm 1936; Jones 1937; ASM 1946; Pilling 1948]. The major described the U.S. Steel family of steels, which included USS application was, as indicated earlier, railroad coal hopper cars Cor-Ten steel. and coal barges. In 1948, an unpainted Cor-Ten steel angle Each alloying metal addition makes its own unique contri- member was installed in a galvanized carbon steel transmission bution to both corrosion resistance and strength. Larrabee and tower in the Gary Steel Works of the U.S. Steel Corporation. Cobum [1961] showed in a comprehensive study the contri- Periodic inspections indicated that it was performing effectively. bution to corrosion resistance made by the elements copper, Encouraged by this example, a bare steel transmission tower phosphorus, silicon, nickel, and chromium. They demonstrated line of Cor-Ten steel was built along Lake Michigan in the Gary the interplay of the various combinations by testing 270 steels Steel Works. An experimental tower also was erected by General with three variations of chromium, five of copper, three of Electric in Massachusetts. The first extended transmission line phosphorus, three of silicon, and two of nickel. was erected by Vepco in Virginia followed by the erection of Porter [1981] pointed out that the optimum balance of cor- similar lines in Missouri and Georgia. 14

The most prominent competing steel was trade marked May- strength and plate thickness increased and the transition tem- ari R (Bethlehem Steel Corporation), based on the fiery ap- perature was raised. pearance of the natural nickel-chromium ore obtained from the The ensuing broad-scale research in the United States, Can- Mayari District in northeastern Cuba. The "R" stood for en- ada, Europe, and Japan on grain size effects, strengthening hanced atmospheric corrosion resistance. Youngstown Steel dis- mechanisms, composition, and steel cleanliness resulted in the tributed its YdliOy brand of atmospheric corrosion-resistant currently available tough and weldable HSLA steels. These steel steel. products combine carefully controlled heating and rolling pro- Many foreign steel producers, as well as domestic producers cedures with optimum use of alloying elements [Porter 1981]. wishing to take advantage of both the well-known Cor-Ten steel In 1968, specification A588 was issued for a high-strength name as well as the enormous amount of production know-how low-alloy steel with 50-ksi (345-MPa) minimum yield strength available from U.S. Steel, took out a license for its manufacture for plate up to 4 in. (102 mm) in thickness. Its composition was and sale. It was in the early 1970's that Bethlehem Steel Cor- intended for use in welded bridges and buildings. Its atmospheric poration began referring to its Mayari R steel as "weathering corrosion resistance is approximately equal to twice that of steel." Use by the construction industry of the weathering steel structural carbon steel with copper. It is available in several phrase now makes it recognizable by the general public. proprietary grades each based on minor variations of the same Because of the proprietary nature of the Cor-Ten, Mayari R basic chemical composition. and Yolloy steel compositions, and a number of variations sim- The satisfactory experience with transmission towers and ilarly effective in exhibiting enhanced atmospheric corrosion buildings, coupled with the possibility of avoiding or minimizing resistance beyond that offered by copper-bearing steel, the basic costly maintenance painting of the structures during their service compositions and mechanical properties are listed in the manner lives, led to the construction in 1964-1965 of the first four shown in ASTM Specifications A242, A606, A588, and A709. unpainted weathering steel bridges. Iowa erected a bridge on Specifications A61 8 and Al 14 deal with pipe and tubing. Iowa 28 over Racoon River. The Michigan Highway Depart- Two outstanding and comprehensive papers mark the con- ment built four structures at the crossing of the Eight-Mile road temporary history of the HSLA steels and the efforts to place over U.S. 10 in Detroit. Ohio replaced an old bridge with a new on a firm basis the necessary alloy content, as well as the possible all-welded three-span steel bridge across the Brush Creek in operating mechanism and the influence of widely differing en- Miami County north of Dayton. New Jersey erected a four- vironments [Larrabee 1961; Copson 1960]. Works of this type span composite bridge as an adjunct to the New Jersey Turnpike led to a change in the A242 specification to include the enhanced near Morristown. Since then about 1,800 bridges on state high- atmospheric corrosion resistance and the need for the customer way systems have been built of unpainted weathering steel. This to discuss with the steel supplier his exact application so that figure does not include bridges owned by local government the appropriate steel composition would be supplied. agencies. It is of interest that the first two built, in Iowa and The John Deere Administration Center, located on the out- Michigan, are now scheduled for remedial painting. skirts of Moline, Iffinois, includes a pedestrian bridge leading In 1975, ASTM issued a general specification A709 for Struc- to a two-story exhibition building, which is attached to the seven- tural Steel for Bridges to cover carbon, HSLA and quenched story administration building. Additional buildings have since and tempered alloy steels. Five grades are available in three been added. This complex represents the first instance in which strength levels: 36, 50, and 100 ksi (250, 345, and 690 MPá). unpainted Cor-Ten steel was used for structural members and Grades SOW and 100W exhibit enhanced atmospheric corrosion all exterior columns, posts, beams, girders, and sun control resistance. devices. Prior specifying Cor-Ten steel, a 2-year exposure test The following three sections summarize the evolution of the was conducted to establish the performance of the steel in the specification requirements for the three domestic weathering Moline environment [Coburn 1965]. steels, ASTM designations A242, A588, and A709. The sum- The first application of a weathering type steel in a high-rise maries are broken down into scope, chemical requirements, structure took place in the heart of a metropolitan business tensile requirements, and supplementary requirements. Knowl- district, with the erection of the 32-story Chicago Civil Courts edge of the changes made in the scope and requirements over Building in 1964-1965. It was erected with uncoated steel cur- the years is helpful in assessing the corrosion performance of tain walls consisting of fascia, column covers, and window the older bridges in particular. A summary of some foreign frames. Test panels exposed about 1/2 mile (800 m) from the weathering steel specifications concludes this chapter. building had indicated that the environment was satisfactory and the terminal color would likely be reached in about 3 years. ASTM A242 STEELS All exposed steel was fabricated from A242 Type I (Cor-Ten A) steel. The Picasso sculpture on the plaza was fabricated from The ASTM Standard Specification A242 corresponds to the A588 Grade A (Cor-Ten B) steel [Coburn 1965]. ASHTO Standard No. M161. The group of steels designated as ASTM A588 in 1968 were Since the first publication in 1941, 17 editions of the ASTM part of a broader effort to develop HSLA steels capable of being Standard Specification A242 for high-strength low-alloy struc- welded in thick sections. The development effort began after tural steel have been issued. The current edition is dated 1981 World War II, when concern about the brittle long running fractures experienced in ships, pressure vessels, and such welded Scope structures as bridges focused attention on the fracture toughness and the ductile-to-brittle transition temperature of plate steels. The 1941 edition covered "Low Alloy Structural Steel," suit- Corrosion-resistant HSLA steels could be produced economi- able for welding and riveting, and intended primarily for use as cally by hot rolling, but the weldability generally decreased, as main stress-carrying material of structural members. The ma- 15 terial was limited to thickness not under 3/16 in. (5 mm) and Table 1. Chemical composition of A242 steel, 1941-1983. not over 2 in. (50 mm). No reference was made to type of product and corrosion resistance. Composition, % (Ladle Analysis) In 1942, the scope was expanded to cover structural members where saving in weight and atmospheric corrosion resistance Designation Type C, max. Mn, max. P. max. S. max. Cu, mm. were important. In 1955, the specification title changed to "High-Strength A242-41 - 0.20 1.25 - 0.05 - Low-Alloy Structural Steel." Steel shapes, plates, and bars were A242-55 - 0•22b 1.25 - 0.05 - included in thicknesses up to 4 in. (102 mm). 4242-68 1 0.15 1.00 0.15 0.05 0.20 In 1960, bolted construction was added to the scope. The 2 0.20 1.35 0.04 0.05 0.20' atmospheric corrosion resistance was quantified as being equal to or greater than that of carbon structural steels with copper. If the steel was specified for materially greater atmospheric If chromimn and silicon contents are each 0.50% mm., then the copper corrosion resistance than structural carbon steel with copper, 0.20% mm. requirmnent does not apply. the purchaser was expected to consult with the manufacturer. For composition with a maximum carbon content of 0.15%, the maximmm Welding characteristics were said to vary with type of steel limit for manganese may be increased to 1.40%. furnished. In 1968, concurrently with the adoption of the new A588 specification, several changes were made in the A242 specifications. The atmospheric corrosion resistance was specified as being at least two times that of carbon structural steel with corrosion resistance was raised to at least four times that of copper, which is equivalent to four times carbon structural steel carbon structural steel without copper, the chemical require- without copper (0.02 percent max.). When required, the man- ments were changed in three ways: (1) A242 Type 2 steel with ufacturer had to supply evidence of corrosion resistance satis- low phosphorus content (0.04 percent max.) was added; (2) factory to the purchaser. That left the A242 steels with a higher maximum phosphorus and minimum copper contents were spec- corrosion resistance than the A588 steels, which were specified ified for both types; and (3) if chromium and silicon contents to have atmospheric corrosion resistance approximately (not at of Type 2 steel were each 0.50 percent mm., the 0.20 percent least) four times that of carbon structural steel without copper. mm. copper requirement did not apply. The chemical require- Welding technique being of fundamental importance, the ments for Type 1 and Type 2 are given in Table 1. The elements A242 specification presupposed that welding procedure would commonly added in addition to those listed were chromium, be in accordance with approved methods. nickel, silicon, vanadium, titanium, and zirconium; the first The scope was last changed in 1975, when the atmospheric three mainly for corrosion resistance, the other three mainly for corrosion resistance was reduced from "at least" to "approxi- mechanical properties. Type 2 steel was intended for use when mately" four times that of carbon structural steel without cop- better impact properties were needed. The basic premise re- per. mained unchanged, even to date. That was to let the manufacturer choose and use the alloying elements, in combination with Chemical Requirements those prescribed within limits, that gave the atmospheric car- rosion resistance and mechanical properties. The first edition, A242-41 (note that the date refers to the year in which the specification was published), specified the maximum alloying content of carbon, manganese, and sulfur Tensile Requirements listed in Table 1. The manufacturer was free to use such alloying elements—with carbon, manganese, and sulfur, within pre- Table 2 gives the tensile requirements for A242 steel that were scribed limits—as would give the specified physical properties specified in the original 1941 edition, for material thicknesses and suitability for welding under the given conditions. He had varying from 5/16 in. (8 mm) to 2 in. (51 mm). The basic values to analyze each melt and report the percentages of carbon, of 50-ksi (345-MPa) yield point and 70-ksi (485-MPa) tensile manganese, phosphorus, sulfur, and any alloying elements used. strength at the lowest thickness range are the same as the values The 1942 edition stated the intent that the alloying elements specified today. In 1950, the minimum elongation in 8-in. (200- should be chosen so as to materially increase the atmospheric mm) gage length was reduced to 18, 19, and 20 percent for the corrosion resistance of the steel. three thickness ranges. In 1952, the requirement of 24 percent In 1955, the maximum carbon content was raised to 0.22 minimum elongation in 2-in. (50-mm) gage length was added percent. For compositions with a maximum carbon content of for the thickness range over 1 /2 in. (38 mm) to 2 in. (51 mm). 0.15 percent, the maximum limit for manganese was allowed to In 1955, the yield point and tensile strength at the two higher increase to 1.40 percent. thicknesses were raised to the values that are still specified today, In 1960, the manufacturer had to determine and report the as indicated by the entries for A242-81 in Table 2. The elon- content of all alloying elements found by a ladle analysis for gation in 8 in. (200 mm) was lowered to 19 percent for the information purposes only, to identify the type of steel used. thickness range lX in. to 4 in. (38 mm to 102 mm). This The purchaser had to indicate and consult with the manufac- elongation was further reduced to 16 percent in 1964. Tensile turer if he needed materially greater atmospheric corrosion re- requirements were also added for three groups of structural sistance than that of structural carbon steel with copper. shapes. The values were identical to those for plates and bars In 1968, the same year when the requirement of atmospheric of corresponding thicknesses. 16

Table 2. Original and present tensile requirements for A242 steel.

ASTM Structural Plates & Bars Tensile Yield Elongation in Elongation in Specification Shapes Thickness Strength Point 200 sin (8 in) 50 sin (2 m) m (in) MPa (ksi) MPa (ksl) mm,, B miii., B

A242-41T - 8 < t < 19 485 (70) 345 (50) 21 - (5/8

A242-81 tUg 485 (70) 345 (50) 18 - Type 1&2 - (t<3/4) 19

In 1970, the elongations for all thicknesses of plates and bars, Scope and all groups of structural shapes were set at a uniform 18 percent in 8-in. (200-mm) gage length and 21 percent in 2-in. In 1968, ASTM adopted the Standard Specification A588-68 (50-mm) gage length. With that last change, all tensile require- for "High-Strength Low-Alloy Structural Steel with 50-ksi (345- ments had become identical to those of the A242-8 1 edition MPa) Minimum Yield Point to 4 in. (100 mm) Thick." This listed in Table 2. specification covers high-strength low-alloy structural steel shapes, plates, and bars for welded, riveted, or bolted construction, but intended primarily for use in welded bridges and build- Supplementary Requirements ings where saving in weight or added durability are important. The atmospheric corrosion resistance of A588 steel is approx- The current edition of the A242 specification refers to the imately two times that of carbon structural steel with copper, ASTM A6 specification entitled "General Requirements for which is equivalent to four times carbon structural steel without Rolled Steel Plates, Shapes, Sheet Piling, and Bars for Structural copper (Cu 0.02 percent max.). Welding technique is of fun- Use" for standardized supplementary requirements for use at damental importance. It is presupposed that welding procedure the option of the purchaser. Those considered suitable for use will be suitable for the steel and the intended service. This with the A242 specification are: S2 Product Analysis, S3 Sim- specification is limited to material up to 8 in. (200 mm), inclu- ulated Post-Weld Heat Treatment of Mechanical Test Coupons, sive, in thickness. The scope has remained unchanged in the 12 S5 Charpy V-Notch Impact Test, S6 Drop Weight Test, S8 editions that were issued from 1968 to 1982. Ultrasonic Examination, S14 Bend Test, and S15 Reduction of In 1969, the user was cautioned that compliance with the Area Measurement. The bend test requirement was mandatory A588 standard may require the use of an invention covered by from 1941 to 1974. It became optional in 1975, at the request patent rights. This note was subsequently deleted in 1971. of the purchaser. The other requirements were referenced beginning in 1975 (S2, S3, S5, S6, and S8) and 1979 (S15). Chemical Requirements

ASTM A588 STEELS A588 steel comes in nine grades A to K, excluding I, each proprietary to an individual producer. Table 3 identifies the The ASTM Standard Specification A588 corresponds to the producers of each grade and the corresponding proprietary AASHTO Standard No. M222. names of their steels. In some instances, several producers man- 17 ufacture a specific grade and many producers manufacture more Table 3. Grades, producers, and proprietary names of A588 steels. than one grade of A588 steel. Grades A to G appeared in the original 1968 edition. Grades H, J, and K were added in 1969, Grade Producer Proprietary Name 1970, and 1980, respectively. Grade G was deleted in 1983. The chemical requirements, as specified in the latest A588- A U.S. Steel Cor-Ten B 82 edition, are summarized in Table 4. The minor changes that B Bethlehem Steel Mayan R-50 were made in the compositions of grades A, B, D, and E over the years are given in Table 5. It shows for each grade and CStelco Stelcoloy 50 element what the content was in 1968, when it changed, and 0 Great Lakeo Steel NAX High Tensile what it is today. E Youngstown Sheet & Tube Yolloy High Strength As will be shown in Chapter Three, silicon, nickel, chromium, F Republic Steel Republic 50 and copper contribute most to improving atmospheric corrosion G Armco resistance. Various combinations of these elements provide es- High Strength A sentially the same results. To illustrate the variability of ap- H Kai ser Steel Kai sal 1 oy 50 proaches taken by the 10 producers, Table 6 approximately J Jones & Laughlin Jal -Ten quantifies the content of these elements as low, medium, and K Republic Steel Dura Plate 50 high. These terms are relative to the range of content in all

Table 4. Present chemical composition of A588 steels.

Designation Grade Composition. %

C Mn P S Si Ni Cr Mo Cu V Zr th Ti max. max.

A .19 .80 .04 .05 .30 .40 .40 .25 .02 max. 1.25 .65 max. .65 .40 .10 B .20 .75 .04 .05 .15 .50 .40 .20 .01 max. 1.25 .50 max. .70 .40 .10 C .15 .80 .04 .05 .15 .25 .30 .20 .01 max. 1.35 .30 .50 .50 .50 .10 D .10 .75 .04 .05 .50 .50 .30 .05 .04 .20 1.25 .90 .90 max. .15 max. E .15 1.20 .04 .05 .30 .75 .08 .50 .05 max. max. max. 1.25 .25 .80 sax.

F .10 .50 .04 .05 .30 .40 .30 .10 .30 .01 .20 1.00 max. 1.10 max. .20 1.00 .10 G .20 1.20 .04 .05 .25 .80 .50 .10 .30 .07 max. max. .70 max. 1.00 max. .50 max.

H .20 1.25 .035 .040 .25 .30 .10 .15 .20 .02 .005 max. max. .75 .60 .25 max. .35 .10 .030 J .17 .60- .04 .05 .30 .50 .30 .030 max. 1.00 .50 .70 mm. .050 K .20 .50 .04 .05 .25 .40 .40 .10 .30 .005 max. 1.20 .50 max. .70 max. .50 .05

Table 5. Changes in Chemical requirements of A588 steels, 1968-1983. Table 6. Relative contents of alloying elements in A588 steels.

Grade Year of Element Previous New Element Grades Change Content, % Content, S A,B,K C 0 E F G H J

A 1977a C 0.10 - 0.19 - 0.19 max. Mn 0.90 - 1.25 0.80 - 1.25 Si med low high low low med med med Si 0.15 - 0.30 0.30 - 0.65 Ni none 0.40 max. Ni low low - high high med med med B 1970a C 0.10 - 0.20 0.20 max. Cr med med high - low high low - 1977a Ni 0.25 - 0.50 0.50 max. 1980a M, 0.75 - 1.25 0.75 - 1.35 Cu med med low high high med med high Si 0.15 - 0.30 0.15 - 0.50

0 1977a Cr 0.50 - 0.75 0.50 - 0.90

E 1977a Si 0.15 - 0.30 0.30 max. MD 0.10 - 0.25 0.08 - 0.25 18

Table 7. Tensile requirements for all grades of A588 steel.

Structural Plates & Bars Tensile Yield Elongation in Elongation in Shapes Thickness Strength Point 200 mi (8 in) 50 nun (2 in)

nun (In.) MPa (ksi) MPa (ksi) miii., S miii., S

t<102 (E<4) 485 (70) 345 (50) 18 21

102< t< 127 (4

12 7< t< 204 (S

All groups 485 (70) 345 (50) 18 21

grades. For example, Grades A, B, and K combine low nickel plates intended for use in bridges. The five available Grades 36, with medium silicon, chromium, and copper contents. The larg- 50, 50W, 100, and 100W are equivalent to A36, A572, A588, est contrast is apparent by comparing Grades D and E. Grade and A5 14, respectively. Grade 100W corresponds to AS 14 with D is low in copper and high in silicon and chromium; it has atmospheric corrosion resistance. When the supplementary re- no nickel. Grade E is low in silicon, and high in nickel and quirements are specified, the steels exceed the requirements of copper; it has no chromium. Specifications A36, A572, A588, and A514. Specification A709 It should be kept in mind that these elements also influence, conveniently allows the purchaser to specify a bridge steel with in combination with others, the tensile and impact properties the desired tensile requirements, impact test requirements for of the steel, not just the corrosion resistance. The producers several temperature zones, and atmospheric corrosion resistance. balance the contents of all alloying elements to achieve the There are no grade distinctions based on proprietary chemical desired overall behavior. compositions as is the case with Specifications A588 and AS 14. The temperature zones are those chosen by AASHTO and are in relationship to the lowest ambient temperature expected for Tensile Requirements that area. Specification A709 has not changed significantly since 1974; however, a revised version to include complete chemical Table 7 gives the current tensile requirements for A588 steel. analysis requirements, including proprietary grades in A588 and The tensile strength and yield point are the same as those spec- A514, is being considered by ASTM Subcommittee A01.02 on ified in 1968, with one exception. In 1974, the tensile strength Structural Steels. and yield point of Group 5 structural shapes were increased Steel grades with suffix "W" provide atmospheric corrosion from 67 ksi (460 MPa) and 46 ksi (315 MPa), respectively, to resistance about two times that of carbon structural steel with the uniform values of 70 ksi (485 MPa) and 50 ksi (345 MPa) copper, which is equivalent to four times carbon structural steel applicable today for all groups. In addition, the elongation of without copper (Cu 0.02 percent max.). When required, the plates and bars in 8 in. (200 mm) was lowered from the original manufacturer must supply evidence of corrosion resistance sat- 19 percent to 18 percent; and the elongation of structural shapes isfactory to the purchaser. in 2 in. (50 mm) was raised from the original 19 percent to 21 percent. Chemical Requirements

Supplementary Requirements Table 8 gives the basic chemical requirements for carbon, manganese, phosphorus, and sulfur of Grades 50W and 100W. Similarly to Specification A242, the bend test requirement in Specification A709 contains some additional requirements for Specification A588 was mandatory from the first edition to 1974. grain-refining elements (aluminum, columbium, vanadium, and Thereafter it became optional at the request of the purchaser. The additional standardized requirements, for use at the option of the purchaser, were adopted in 1977 (S2, S3, S5, S, and S14) and in 1979 (S15 and S18), all of which are listed in Specification A6.

Table 8. Chemical composition of A709 steels with enhanced atmo- ASTM A709 STEELS spheric corrosion resistance.

Scope Thickness Composition, S (ladleAnalysis) Grade m (iii.) C lii P, max. S. max. In 1974, ASTM issued Standard Specification A709 for "Structural Steel for Bridges." This specification covers carbon 50W 102 (4) 0.20 max. 1.35 max. 0.04 0.05 and high-strength low-alloy steel for structural shapes, plates, 100W 102 (4) 0.10 - 0.21 0.40 - 1.50 0.035 0.04 and bars and quenched and tempered alloy steel for structural 19 titanium); as well as carbon, manganese and silicon depending Supplementary Requirements on material thickness, toughness, and treatment. The choice and use of alloying elements, combined with carbon, manganese, The purchaser may specify supplementary requirements re- phosphorus, and sulfur within the limits prescribed to conform garding fine grain practice, Charpy V-notch impact energy ab- to the tensile and mechanical properties or to enhance the at- sorbed, supplying evidence of atmospheric corrosion resistance, mospheric corrosion resistance, or both, may vary by manufac- and ultrasonic examination. turer.

FOREIGN WEATHERING STEELS Tensile Requirements Tables 10 and 11, respectively, summarize the chemical and The tensile requirements for Grades 50W and 100W, given tensile requirements stated in the weathering steel specifications in Table 9, are the same as those for A588 and A514 steels. issued by Canada [CSA 1976], United Kingdom [BS 1969],

Table 9. Tensile requirements for A709 steels with enhanced atmospheric corrosion resistance.

Brinel 1 Plate Structural Yield Point! Tensile Minimiin Elongation, S Reduction Hardness Grade Thickness Shapes Strength, mm. Strength, in 200 rn in 50 mn of area Nunber

sin (in) MPa (ksi) MPa (ksi) (8 in) (2 in) mm., (5)

50W t<102 (4) Groups 1-5 345 (50) 485 (70) mis. 18 21 - -

100W t<64(2-1!2) - 690 (100) 760-895 (110-130) - 18 40-50 285-293 100W 64

Notes:

The elements colmuobiun and vanadiumi may be used singly or in combination up to the total of 0.10% max., except that if columibiun is used Singly or in combination with vanadiun in plates thicker than 12 xiii or shapes heavier than Group 1, the silicon content shall be 0.15% mis. Almuninum may be used as a grain refining element without prior approval by the purchaser, and when so used, shall not be included in the sumnation of grain refining elements.

The combined content of chromium, nickel and copper shall be not less than 1.00%.

The combined content of chromiun and nickel shall be not less than 0.40%.

At the purchaser's request or at the producer's option, the steel may be made with no minimuim silicon content provided that the steel contains a mirmimumi of 0.02% total aluninun content.

WR 508 may also be supplied in normalized condition.

Steels of Type 4, 33 and 84 shall contain at least one of the following grain refining elements in the indicated percentages: Al total > 0.015%, Nb = 0.01 - 0.060%, V = 0.02 - 0.15% and Ti = 0.02 - 0.10%.

The steel shall contain nitrogen oxidizing elements in sufficient content to refine the grains.

Content to be agreed upon.

KT 45-2 steel shall be killed with aluninun, maintaining the relationship Al/3 + V/4 > N.

One or more elements aimong Mo, Cb, Ti, V and Zr shall be added. The total content of these elements shall not exceed 0.15 percent.

Table 10. Chemical requirements for foreign weathering steels for bridges.

Type Composition, N or Grade C Mn P S Si Ni Cr Mi Cu V Zr Cb Ti Al N Connnents Canada: G40.21 [CSA 1976] 070C 50A .20 .75 .03 .04 .15 0.90c .20 a a a Killed; fine max 1.35 max max max .90 .70 .60 arain oractice max max .lO 60A .20 .75 .03 .04 .15 .90c .20 a a a Killed; fine max 1.35 max max .40 max max .60 grain practice

United Kingdom: BS 4360 [BS 1969]

WR50 .10 .90 .04 .05 .15 .40 .25 .02 .01 As rolled; killed, B&B1 .19 1.25 max max .50 .70 .40 .10 .06 e

WR50C .10 .90 .04 .05 .15 .40 .25 .02 .01 As rolled or nor- .22 1.45 max max .50 .70 .40 .10 .06 malized; killed

WR50C1 .10 .90 .04 .05 .15 .40 .25 .02 .01 As rolled or nor- .19 1.25 max max .50 .70 .40 .10 .06 malized; killed France: A35-502 [AFNOR 1979]

E24W .13 .20 .04 .035 .10 .65 .40 .20 f 2,3&4 max .60 .40 max .80 .55

E36W .19 .50 .04 .04 .50 .65 .40 .30 .20 .15 B3&B4 max 1.50 max max max max .80 max .55 max Federal Republic of Germany: DASt 007 [DASt 1979]

WT St- .13 .20 .05 .b35 .10 .50 .30 0.007 37-2 max .50 max max .40 .80 .50 WT St- .13 .20 .045 .035 .10 .50 .30 0009 9 37-3 max .50 max max .40 .80 .50 WT St- .13 .90 .045 .035 .10 .50 .30 .02 0.009 9 52-3 max 1.30 max max .50 .80 .50 .10 German Danocratic Republic: TGL 28192 [DDR 1973] 03h i KT 45 .12 .30 .05 .04 .25 .50 .30 0.02 0.008 2 & 3 max .60 .09 max .50 .80 .50 min 0.016 KT 52-3 .12 h .05 .04 .25 .50 .30 h h h max .09 max .50 .80 .50 Czechoslovakia: [CSSR 1978]

Atmofix .12 .20 .04 .04 .10 .20 .30 .30 37 max .55 max max .35 .40 .60 .50

Atmofix .10 .90 .04 .04 .20 .30 .40 .30 .02 .02 .01 528 .17 1.20 max max .45 .60 .80 .55 .06 .06 mm

Atmofix .16 1.10 .04 .04 .20 .30 .60 .30 .03 .01 60 max 1.50 max max .40 .50 1.00 .50 .08 mm Japan: JIS G3114 (Non-painted application)3 [JIS 1981]

SMA 41 .18 1.25 .035 .035 .15 .05 .45 .30 A,B&C max max max max .65 .30 .75 .50

SMA 50 .18 1.40 .035 .035 .15 .05 .45 .30 A,B&C max max max max .65 .30 .75 .50 SMA 58

Japan: JIS G3114 (Painted application)3 [JIS 1981]

SMA 41 .18 1.25 .035 .035 .55 .30 20 A,B&C max max max max max .55 35

SMA 50 .18 1.40 .035 .035 .55 .30 20 A,B&C max max max max max .55 35 SMA 58

France [AFNOR 1979], Federal Republic of Germany [DASt other than type of steel. Only types/grades of weathering steels 1979], German Democratic Republic [DDR 1973], Czechoslo- that appear to be used for bridges were included. Among the vakia [CSSR 1978], and Japan [JIS 1981]. This information is steels that were deleted are those with maximum phosphorus provided to show that (1) domestic and foreign steels are similar content greater than 0.10 percent and those supplied in plate in composition and strength; and (2) any differences in corrosion thicknesses of Y2 in. (12 mm) or less. These types/grades, like resistance and fatigue behavior, examined in Chapters Three ASTM A242 Type 1, are mainly used for architectural appli- through Five, Seven, and Eight, are mainly a result of factors cations. 21

Table 11. Basic tensile requirements for foreign weathering steels for bridges.

Type Plates & Bars Tensile Yield Max. or Structural Thickness Strength Point Thicknes1 Grade Shapes m (in.) MPa (ksi) I5'a (ksi) Available

Canada: G40.21 [CSA 1976]

50A All groups t<102 480 - 660 345 102 t(4) (70 - 95) (50) (4) 60A Groups 142 t<38 520 - 695 415 38 t7.(1-l/2) (75 - 100) (60) (1-1/2)

United Kingdom: BS 4360 [BS 1969]

WR50 t<40 480 345 50 8,B1,C&C1 t(1.1/2) (70) (50) (2)

France: A35-502 [ADNOR 1979]

E24W 1:<30 3

German Democratic Republic: TGL 28192 [DOR 1973]

KT 45 t<16 440 - 590 295 40 243 t(5/8) (64- 85) (42) (1-1/2)

KT 50-2 t<16 490 - 640 345 16 tZ(5/8) (70 - 92) (5/8) KT 50-3 t<16 510 - 610 355 40 t7(5/8) (74 - 88) (1-1/2)

Japan: JIS G3114 [JIS 1981]

SMA 41 t(16 420 - 539 245 A,B,&C (1<5/8) (58 - 78) (36) 1640 420 - 539 216 (t>1-9/16) (58 - 78) (31)

SMA 50 t<16 490 - 608 363 A,B.&C (1<5/8) (71 - 88) (53) 1640 490 - 608 333 (t>1-9/16) (71 - 88) (48) SMA 58 t<16 569 - 716 416 (1<518) (83 - 104) (67) 1640 569 - 716 431 (t>1-9/16) (83 - 104) (63)

Note:

(a) Tensile strength and yield point may be lower when the material thickness exceeds the value listed in the third colunn of the table.

The various type/grade designations identify steels with different levels of tensile requirements. Each country has a weathering steel with basic tensile strength and yield point, Domestic and foreign weathering steels have similar contents respectively, comparable to the values of 70 ksi (485 MPa) and of alloying elements added mainly for improving corrosion re- 50 ksi (345 MPa) for ASTM A242 and A588 steels. Some have sistanëe. Most have, on average, 0.30 to 0.50 percent copper, in addition higher and lower grades, with yield points ranging 0.04 percent max. phosphorus, and greatly varying contents of from 34 ksi (235 MPa) [AFNOR 1979] to 67 ksi (461 MPa) nickel. chromium, and silicon. As will be shown in Chapter [JIS 1981]. Listed in Table 11 for reasons of brevity are only Three, a desired level of onrrosion resistance can be achieved the tensile requirements for the smallest material thickness. The with many different combinations of these elements. requirements are usually lower for thicker plates and sections. Like the ASTM standards, the Canadian standard specifies The type/grade designation implies in several cases a mini- that the corrosion resistance of weathering steel is approximately mum level of Charpy V-notch energy absorbed at specified four times that of carbon steel (0.02 percent Cu). The other service temperatures. These values are stated in the correspond- foreign standards examined herein do not quantitatively specify ing references. the corrosion resistance. 22

CHAPTER TWO

EXPERIENCES

DOMESTIC USE Dakota question the corrosion performance, particularly in salt environments, and dislike the rust staining of the concrete sub- The extent to which weathering steel (unless noted otherwise, structure. South Carolina takes the cautious approach of not the term weathering steel should be interpreted as being in the adopting it as long as others are having problems. unpainted condition) is used for bridges on state highway sys- Many steel users, including those who have rejected bare steel tems was determined from telephone interviews with officials construction, have specified for economic reasons painted A588 in all 50 States and the District of Columbia. Table 12 sum- steel either outright or as substitute for A572 in thicknesses marines the findings. In most cases the fmdings are the same greater than 2 m. (50 mm). This is done when the savings in as those obtained in separate surveys by Bethlehem Steel and weight is greater than the additional cost of high-strength low- the AASHTO Subcommittee on Maintenance [Bethlehem 1982, alloy steel, as compared to A36 steel. AASHTO 1983]. According to the present survey, there are about 1,800 weathering steel bridges on state highway systems, not including DOMESTIC EXPERIENCE county, city, tollroad or mass transit bridges. Forty-three of 50 States and the District of Columbia have weathering steel The majority of bare weathering steel bridges built in the bridges. Of those, 31 states and the District of Columbia foresee United States are performing in a satisfactory manner. But the using bare weathering steel for new bridges, 2 states are hesitant, application of atmospheric corrosion resistant steels has not been without problems. and 10 states have stopped using it. Indiana, Iowa, and Michigan have official policies banning the use of unpainted weathering steel for bridges on the state highway systems. There are several Michigan satisfied users among the 31 states and the District of Columbia who continue to build weathering steel bridges. Vermont, for During a lengthy evaluation period, which began with in- example, specifies it almost exclusively for steel bridges. Many spections of Detroit's Eight-Mile Road Bridge, Michigan DOT states carefully select the sites, avoiding industrial environments officials found that corrosion rates were not tapering off. As a with severe atmospheric pollution, marine environments along result, the State of Michigan instituted in June 1979 a partial the seashore and water ways, urban areas with high deicing salt moratorium on the use of A588 steel in the unpainted condition use, and inland sites with high humidity and rainfall. Some for bridges in the following situations [VanKampen 1979]: states paint the steel girders up to 10 ft (3 m) on either side of joints. For aesthetic reasons Illinois, Missouri, and Wisconsin Depressed roadway sections where low underclearance use weathering steel mainly for stream crossings; Wisconsin, (less than 20 ft (6.1 m)) and vertical retaining walls trap salt also for grade separations in remote areas. The overriding reason sprays and other atmospheric pollutants. why these States continue to specify weathering steel is the Urban and industrial areas where heavy roadway salting potential of saving the cost of maintenance painting. By virtue and automotive and industrial pollution create an aggressively of having generally favorable environments or by carefully ap- corrosive environment. plying the material, they are able to make the concept of bare weathering steel construction work. Nine months later, in March 1980, the moratorium was ex- Of the 10 states who have stopped using weathering steel, tended to include all uses of unpainted A588 steel on the state Alabama, Indiana, Iowa, Michigan, Mississippi, and West Vir- highway system [VanKampen 1980]; This ban included rural ginia have done so because of corrosion problems encountered bridges not affected by the earlier moratorium. The fmdings in roadway deicing salt environments. Alaska discontinued the and conclusions that led to the moratoria were thoroughly doc- use because of the high cost of steel and a generally humid umented and explained [Culp 1980, Arnold 1980, Allemeier environment along its southeastern coast. The long life of paint- 1981]. The following forms of corrosion were observed: ing systems in the dry climates of Montana makes painted steel construction more economical than bare weathering steel con- 1. By far the most serious problem is salt contamination in struction. In North Dakota concrete bridges are more econom- the following manner: (a) runoff water drains onto the super- ical than steel bridges. structure through expansion joints and leaky seals; and (b) spray That leaves 7 states who have never used bare weathering kicked up by traffic passing under the bridge accumulates on steel. Their main reasons are as follows. Arizona, Nevada, and the superstructure. Corrosion attack by leakage is more apparent New Mexico have dry climates in which painting systems last because it is concentrated, and the steel area affected is a small indefmitely and weathering steels would not for a long time percentage of the total area. It is prevalent below the leaks, but develop an oxide coating of uniform appearance. Steel is ex- the salt water can advance for long distances along the bottom pensive in Hawaii, relative to concrete. New Jersey and South flange, both up and down grade from the source. Corrosion 23

Table 12. Summary of use of bare weathering steel bridges on federal highway systems.

Weathering Steel Bridges in Service New Weathering Steel Bridges

Previously No. of Used Future State Used Bridges Since Consent Use Consent

Alabama Yes 5 1980 -- No Concerned about experiences by others.

Alaska Yes 2 1973 -- No Concrete is more economic; unsuitable for humid SE coastal area.

Arizona No -- Dry climate; paint No (Same) systems last indefi. nitely; nonuniform weathering.

Arkansas Yes 50 1970 -- Yes Used mainly where maintenance is difficult

California Yes 1978 -- Yes Careful site selection.

Colorado Yes 25 1975 -- Yes --

Connecticut Yes 40 1970 -- Yes Used mainly where maintenance is difficult.

Delaware Yes 10 1975 -- Yes --

District of Columbia Yes 3 1978 -- Yes --

Florida Yes 2 1972 12 painted W/S No Unattractive. bridges.

Georgia Yes 70 1976 Used in nonhostile Yes -- environments.

Hawaii No -- Last steel bridge No Concrete is more built in 1965. economic.

Idaho Yes 15 1970 Not used near smelt- Yes (Same) ers and processing plants.

Illinois Yes 31 1968 Mainly rural bridges Yes (Same) over streams. Indiana Yes 10 1970 -- No Policy based on Michigan experience.

Iowa Yes 5 1964 -- No Policy.

Kansas Yes 22 1971 Mainly painted hybrid Yes -- girders in urban areas.

Kentucky Yes 2 1971 -- Yes --

Louisiana Yes 16 1975 -- Yes

Maine Yes 60 1968 -- Yes --

Maryland Yes 66 1973 -- Yes --

Massachusetts Yes 20 1972 Not used over stream Yes (Same) crossings.

Michigan Yes 480 1965 -- No Policy; severe corrosion effect of deicing salts. Minnesota Yes 60 1973 -- Yes Painted in metropolitan bare in rural areas.

Mississippi Yes 16 1978 -- No Salt usage in ice storms. Missouri Yes 20 1973 Mainly rural bridges Yes (Same) over streams.

Nontana Yes 2 1973 Initially nonuniform No Dry climate; paint weathering, systems last 30 years.

Nebraska Yes 22 1969 -- Yes -- Nevada No -- Dry climate; non- No (Same) uniform oxide coating; esthetics.

30. New Hampshire Yes 100 1973 -- Yes -- 24

Table 12. Continued

Weathering Steel Bridges in Service New Weathering Steel Bridges

Previously No. of Used Future State Used Bridges Since Conuiient Use Convent

31. New Jersey No -- -- Concern about loss No (Same) of section due to corrosion.

32. New Noxico No -- -- Dry climate; paint No (Same) systems have long life.

33. New York Yes 70 1969 -- Yes --

34. North Carolina Yes 150 1968 -- Yes --

35. North Dakota Yes 3 1970 Semi-arid area; No Not economic compared snow is mostly plowed, with concrete.

36. Ohio Yes 55 1971 -- Yes

37. Oklahoma Yes 4 1973 9nall jobs bare; Under -- large jobs painted. evaluation

38. Oregon Yes 3 1975 -- Yes --

39. Pennsylvania Yes 12 1968 -- Yes Except in polluted and hwiid areas. Difficult to inspect.

40. Rhode Island Yes 9 1975 -- Yes --

41. South Carolina No -- -- Esthetics; problems No (Same) in other states

42. South Dakota No -- -- Esthetics; experience No (Same) in other states with deicing salts.

43. Tennessee Yes 10 1974 -- Under Esthetics eval uation

44. Texas Yes 16 1974 -- Yes --

45. Utah Yes 14 1970 -- Yes --

46. Vermont Yes 100 1973 Used extensively Yes (Same)

47. Virginia Yes 45 1972 -- Yes --

48. Washington Yes 15 1967 -- Yes --

49. West Virginia Yes 10 197Z No Experience in other states

50. Wisconsin Yes 25 1971 For stream crossings & Yes (Same) grade separations in remote areas.

51. Wyoming Yes 20 1975 Dry climate Yes -- attack by spray can have just as great an effect as corrosion by contaminated with salt. This raises a concern about the fatigue leakage. It covers most steel surfaces. It is most severe over the resistance which is known to be reduced by pitting. traffic lanes and when the roadway is depressed and laterally The electrical potential difference between the mill scale confmed by retaining walls. and the base metal selectively corrodes the base metal adjacent Corrosion products, salt, dust, bird droppings, and othet to the mill scale. This galvanic corrosion caused widespread debris continuously accumulate, especially on top of shelterØ pitting on some bridges that were exposed to a rural eaviron- bottom flanges. They trap moisture and create a corrosive poul- ment, without heavy exposure to salt. tice. Rain can only wash off debris on the outside surfaces of the fascia girders. The protective oxide coating does not form where the steel The capillaries in the oxide coating draw water and dis- is severely corroding. The flaky and laminar rust is periodically solved deicing salts several inches up the web, corroding the shed. Under these adverse conditions unpainted A588 steel is lower web strip along the flange-to-web junction more than the corroding at about the same rate as unpainted carbon steel. The upper web surfaces. Michigan DOT is now retrofitting the steel bridges that are in The narrow gaps between loosely matched hanger plates, the worst condition. This includes replacement of joint seals pin connections, and attachments foster crevice corrosion, an and maintenance painting of girders. This rehabilitation pro- accelerated form of attack. The pressure of the corrosion prod- gram is expected to last many years since nearly one-fourth of ucts froze some hinges and failed pin nuts. all weathering steel bridges on the federal highway system are The surfaces of A588 steel were pitted, especially when located in Michigan. 25

AISI Study officials found significant corrosion of the steel. It was most prominent on top of the bottom flange, on the exterior portion The Michigan ban focused general attention on the perform- of the seaward girder. This part of the flange was generally ance of weathering steel bridges. Under the auspices of the covered with a Y, to %-in. (6 to 10-mm) thick layer, of debris American Iron and Steel Institute, a task group of steel company consisting of a mixture of dirt and rust scale. The rust-scale specialists and state and federal highway officials inspected 49 represented the major portion of the debris and had fallen off weathering steel bridges in Illinois, Maryland, Michigan, New the lower 2 to 3 ft (0.6 to 0.9 m) of the web. The corrosion was York, North Carolina, Wisconsin, and New Jersey Turnpike most severe near the web-flange junction. At one location, a Authority [AISI 1982]. They found that: 4 x 8 x Y16 (100 )( 200 x 1.6 m) thick rust sheet was removed intact from the flange. Rust pieces of similar thickness Thirty percent of the bridges showed good performance could also be removed from the surface of the web-flange fillet in all areas. weld. The bridges were also corroding significantly at locations Fifty-eight percent of the bridges showed good overall free of accumulated debris. All bridges are within Y, mile (400 performance with moderate corrosion in some areas. m) of salt water, and the area receives 150 to 180 in. (3.8 to Twelve percent of the bridges showed good overall per- 4.6 m) of annual rainfall. formance with heavy local corrosion in some areas. The following fmdings were based on metallography, scanning electron microscopy, X-ray elemental analysis, and ultrasonic The inspection reports indicated that the major portion of thickness measurements performed by a consultant [Kasza the steel had developed the expected oxide coating. One bridge 1983]: on NY 8 with minimum clearance over the polluted Nine Mile Creek, near Syracuse, and several Michigan bridges did not develop the expected oxide coating. Other bridges exhibited The composition of the steel was within the range specified moderate to heavy corrosion near leaky expansion joints in the by ASTM for A588 Grade A, with the exception of a slight bridge deck. The factors found to be responsible for the for- reduction in silicon which is unlikely to affect the corrosion mation of the nonadherent flaky rust are the same as those performance. described in the earlier Michigan report [Culp 1980]. The scale was a corrosion product of the base metal, not Most bridges, that were inspected did, not need immediate mill scale as had been suggested. attention or overall painting. Notable exceptions are most of The corrosion products were primarily crystalline iron ox- the structures inspected in the Detroit, Michigan, area [AISI ide which is nonprotective. They contained only traces of chlo- 1982]. ride and sulfur. Absence of significant levels of those corrosive The high rate of salt application in Wayne County and De- elements indicates that the prolonged time of wetness, without troit, the widespread use of cantilevered/suspended spans, and adequate drying time, was the main reason for the rapid cor- the large number of weathering steel bridges have resulted in a rosion. greater number of known corrosion incidents in Michigan than The steel was corroding at an estimated rate of 2 mils (50 in other states. m) per year and per surface. That is an order of magnitude However, the problems arising from unsuitable applications higher than the normal corrosion rate expected of weathering of weathering steel are not limited to Michigan alone. There steels with protective oxide coatings. The total section loss could are known cases of weathering steel bridges in Alaska, Califor- not be determined because the initial thickness of the steel had nia, Indiana, Louisiana, and Ohio that have not performed to not been measured. After 10 years of exposure, the section loss expectation. Some of them have remained practically unknown below nominal design values was too small to affect the load to the engineering profession. For example, the weathering steel carrying capacity (0.013-in. (0.33-mm) maximum loss in thick- bridge on County Road 37 over Little Still Water Creek, Tus- ness of nominally 0.563-in. (14.3-mm) thick web). carawas County, Ohio, was built in 1973 and had to be painted in 1979, before the Michigan DOT reported on the condition The high annual rainfall and the proximity to bodies of water of their weathering steel bridges [Smith 1983]. Information on create a humid environment along the southeastern coast of th bridges in the aforementioned States are summarized in the Alaska. Wind-driven rain, fog, mist, and possibly water leaking following to illustrate the conditions unfavorable to the devel- through the deck frequently wet the steel. In addition, in humid opment of a protective oxide coating. These cases were not environments steel surfaces condense water during the night reported in Ref. AISI 1982. when the temperature drops by, say, 15 F (8C). This greatly extends the time of wetness. The condensed water slowly runs down the web. Because the web dries from the top down, the Alaska lower web stays wet longer and corrodes more. The accumulated debris on the lower flange traps moisture and aggravates flange In 1972, four A588 Grade A steel bridges were built on the corrosion. Under these conditions, weathering steels corrode at Big Salt Lake Road (FAS 929) east of Kiawock, Prince of Wales a rate comparable to that of unpainted carbon steel. The four Island, Alaska. They span across Little Salt Creek, Duke Creek, Alaska bridges are scheduled for remedial painting in 1984. Black Bear Creek, and Steelhead River. They consist of two This case study is important because it is well documented longitudinal plate girders with transverse floor beams resting [Kasza 1983]. It shows that weathering steel does not develop on the top girder flanges. The floors are longitudinally laminated a protective oxide coating if the time of wetness is long. Salt treated timber. contamination was not a significant factor as it had been in An inspection of the bridges in 1981 by Alaska and FHWA Michigan. 26

California 1975. It was decided in 1976 to inspect these bridges on an annual basis, for a minimum of 10 years, to ensure that main- Five A588 Grades A and H steel bridges in Redwood National tenance painting could be elimirated and that no additional Park, California, have superstructures consisting of rolled beams maintenance problems, unknown at that time, would be inad- and 2 X 8 in. (50 )< 200 mm) or 2 X 10 in. (50 X 250 mm) vertently created. In 1983, 8 years later, it appears that 13 nail-laminated timber decks. One was built in 1973 and four in bridges were performing as expected, whereas 3 bridges were 1975. The first (Prairie Creek Bridge) is located more in the not. The Doullut Canal Bridge, located 40 miles (64 km) S-SE open, appears to be more often dry, and is performing well. The of New Orleans, was remedially painted in 1983. The Larose other four are on Lost Man Creek, in a different drainage on Bridge, 30 miles (48 km) S-SW of New Orleans, is showing the east side of the valley, which is much wetter and has a dense signs of unstable corrosion. Sections of the newly erected Luling flora more like that of a rain forest. These environmental con- Bridge, 18 miles (29 km) W of New Orleans, were at first severely ditions are similar to those found in Southeast Alaska, except corroding because of sea-water contamination during shipping that the Redwood Park bridges are 10 to 15 miles (16 to 24 from Japan [Garrido 1983, Dunn 1983]. km) from the ocean, whereas those in Alaska were within sight The Doullut Canal Bridge, on Louisiana Route 23 in Empire, of the ocean. is located in the lower Mississippi River delta several miles from An inspection of the bridges by FHWA officials in 1981 the Gulf of Mexico and near bodies of brackish water. The five- revealed that the interior girders and the inside face of the span plate girder bridge, made of A588 Grade A steel, has a exterior girders of two bridges on Lost Man Creek were severely 55-ft (16.8-rn) clearance over the Doullut Canal. corroding [Kasza 1983]. Large pieces of rust separated from the The boldly exposed surfaces of the northeast exterior girder top flange and large rust flakes fell off the web. Debris was exhibited a tight oxide of normal appearance. This surface is accumulating along the full length of the lower flange, about '2 in. (12 mm) deep. sheltered from southerly wind-blown fog and receives drying The exterior face of the girders could not be inspected because heat from the morning sun. All other steel surfaces were covered it is clad with slabs of redwood logs to make the structure look with flaky rust. The flakes ranged in size from X x Y. in. (3 x like a log stringer bridge. The humidity trapped in the logs is 3 mm) to Y. X /2 in. (6 x 12 mm). The steel surface beneath certain to foster corrosion of the underlying steel. the rust was covered with broad, shallow pits. Wet poultices of Metallurgical examinations similar to those performed on the rust flakes were accumulating on the top of the lower flanges Alaska bridges showed that: (1) the rust scale consisted of non- and on horizontal gusset plates of lateral bracing members. At protective iron oxides of the base metal, not mill scale; (2) there both locations, nuts partially covered by accumulated rust-flakes was no evidence of any significant concentrations of chlorides were scaling. The threads and the bolts ends protruding from and sulfides; and (3) the corrosion rate was about 2.5 mils (75 some nuts had corroded away. The chloride content of rust g.m) per year and per surface. flakes (0.09 and 0.22 percent) and poultice samples (0.11 per- It appears that frequent rainfall, fog and high humidity are cent) indicated significant chloride contamination. not letting the steel dry. The long time of wetness is fostering The adherent oxide coating had not formed on the majority corrosion at a rate comparable to that of carbon steel. of the surfaces of the DoullutCanal Bridge because of location- related salt contamination of the steel, high incidence of fog, and high humidity. The bridge was sand blasted and painted Iowa (see Ch. Nine) in 1983 at a cost of $195,400 for 17,075 U.S. tons (15,200 metric tons) of steel with 214,500 ft2 (19,950 m2) Four of the five weathering steel bridges on the state highway surface area. system were built between 1971 and 1982. Inspection of these The Larose Bridge on Louisiana Route 657 over an Intra- bridges in 1983 by state highway officials showed that all four coastal waterway is also located in the Mississippi River Delta, have tight rust of dark brown color and no evidence of unusual about 16 miles (24 km) from the Gulf. Some surfaces of this corrosion [Sheeler 1983]. The chloride content of rust samples A588 Grade B steel bridge are developing a granular flaky rust taken from these four bridges varied from none to 0.11 percent. layer that continuously scales off, leliving a pitted surface. How- The fifth and oldest bridge, built in 1964, is located on Iowa ever, the corrosion is less advanced than that at the Doullut 28 over Racoon River, West DesMoines, in a rural/urban en- Canal Bridge, likely because of its greater distance from the vironment. It has been subjected to heavy salting. Most of the Gulf. Both are 8 years old. steel is in the normal light rust condition, except under expansion The superstructure of the cable-stayed Luling Bridge was joints where the rust is flaking and the steel is pitting. An area fabricated from A588 Grade A weathering steel. The sections of flaky rust also extends along the outside web, 2 to 3 in. (50 were transported by ship from Japan and hosed with water on to 75 mm) above the bottom flange, and runs the full length of arrival in Louisiana, to remove the sea-salt residue that may the outside stringers. There are minor areas of chloride drainage have accumulated. There was no indication of severe corrosion through the deck. Analysis of rust samples indicated 0.56 per- when the steel arrived at the project site. The heavy flaking cent chloride content at an abutment, 0 percent on the web of began to show up after erection had commenced. Not all sections an inside stringer, and 0.52 percent at the flange weld. Chloride were equally affected because some were transported in the holds contents of 0.20 percent are considered to be significant. The of the ship, while others were on the deck and received more Racoon River Bridge is scheduled for remedial painting in 1984. salt. After the corrosion problem became evident, the contact surfaces of the bolted joints of sections that had not yet been Louisiana erected were sand blasted. Also sand blasted were all surfaces exposed to view. Before the expoxy wearing surface was applied, Sixteen weathering steel bridges were built in Louisiana since the steel deck was sand blasted to a near-white condition. It 27 was not always possible to remove all rust from the pits in the Texas deck because the pits were smaller than the sand grains. There is no indication at the present time that the severe initial cor- A weathering steel bridge on State Route 124 over the In- rosion is reoccurring. tracoastal Waterway in High Island, northeast of Galveston, is subjected to conditions similar to those previously described for the Louisiana bridges close to the Gulf of Mexico. The bridge Ohio is located 5 miles (8 km) from salt water in the prevailing wind direction and has an 85-ft (26-rn) clearance. Materials and Tests There are known cases of weathering steel bridges that have Division officials of the State Department of Highways and performed poorly in Tuscarawas, Franklin and Butler Counties, Public Transportation, who have periodically inspected the Ohio. bridges since 1979, reported loose rust flakes on the steel surfaces The bridges on County Road 37 over Little Still Water Creek that range from 6 to Y. in. (1.6 to 3.2 mm) diameter. The steel and on County Road 99 over Sugar Creek, Tuscarawas County, surface under the flakes was covered with porous corrosion consist of 120-ft and 124-ft (36.6-rn and 37.8-rn) long trusses products instead of the dense oxide coating typical of good with wide flange members fabricated from A588 Grade A steel. weathering steel performance. The steel surfaces are pitting. They have about 12- to 15-ft (3.7 to 4.6 rn) clearance above the Observations of the corrosion process of various steels along the creek. The County Road 37 Bridge, built in 1973, had corroded coast show that steel corrodes rapidly until a film of corrosion severely, with the worst locations being along the bottom chord, products builds up. Thereafter, the rate steadily decreases until along the inside floor beams and at the truss joints. County the rust flakes. After flaking, the corrosion cycle starts again. officials stated that 4-in. (6.4-mm) thick rust pieces had come Test coupons were exposed 30 months on a rack mounted on off the bottom chord and some sections had lost 15 percent of the west main pier of the High Island Bridge. They were tensile the steel. The corrosion was attributed mainly to prolonged time tested in 1982. The loss of load carrying capacity, as compared of wetness and insufficient drying of the steel. The bridge was to that of the nonweathered control coupons, was converted to remedially painted in 1979. The County Road 99 Bridge, built equivalent loss in thickness. It was found that the coupons had in 1979, was painted as a precautionary measure 2 months after corroded at a rate of 5.6 mils (142 m) per year and per surface it had been opened to service, to avoid the corrosion problems [Raska 1983]. Some state officials believe that the bridge will found in the County Road 37 Bridge [Smith 1983]. have to be painted and that it would be more cost effective to The three-span multigirder bridge on Brand Road over the paint it before pitting becomes severe. North Fork of Indian Run, Franklin County, was fabricated from A588 Grade A steel. It was built in 1979. An inspection of the bridge in February 1983 revealed that the steel beams Summary were rapidly corroding. Large sheets of rust were flaking off in several areas. Most severely affected were the lower one-third The well-documented Michigan experience and the afore- of the web and the lower flange in all three spans. The interior mentioned corrosion problems with bridges in Alaska, Califor- beams corroded more than the exterior beams. When the bridge nia, Iowa, Louisiana, Ohio, and Texas indicate that weathering was inspected at mid-morning, after the fog had dissipated, steel will not develop a protective oxide coating if the surfaces heavy condensation remained on the steel beams, and much remain wet for extended periods of time or are subjected to salt moisture was found under the rust scale. Chemical analysis of contamination from any source. Under these conditions, weath- rust scale samples removed from the beams revealed only traces ering steel corrodes at a rate comparable to that of carbon steel, of chlorides (0.01 percent) arid sulfates (0.07 percent). The creek as is shown in Chapters Three through Five. water was also found to be low in sulfates (0.18 ppm). Evidently, The aforementioned examples are intended to illustrate the deicing salt and sulfates were not significant factors. The rapid conditions under which weathering steel has not performed as rate of corrosion was caused by moisture present for prolonged expected and should, therefore, not be used. It is well to re- periods of time. The low-lying bridge has only 8-ft (2.4-rn) member that most weathering steel bridges built in favorable clearance over the stream, and the terrain limits sunlight and environments are performing well. air circulation. The bridge was remedially painted in 1983 [Circle 1983]. Butler County, located north of Cincinnati, has remedially painted six of their ten weathering steel bridges in 1983. The EUROPEAN EXPERIENCE Howard Road Bridge, built in 1971 over a stream with 10-ft (3-rn) clearance, was badly corroding. The worst areas were The experiences with bare steel construction in Europe are along the lower part of the interior beams. After the steel had partly reflected in the weathering steel specifications that were been blast cleaned it was found that corrosion had fully pene- issued by various countries. The following samples are indicative trated the 0.39-in. (9.9-mm) thick web of the W18 x 55 beams of the European provisions for the proper application of weath- at several locations. The largest rust hole was about 3 in. by 2 ering steels. in. (75 mm long by 50 mm high). Parts of the beam webs were plated to avoid having to post a weight limit on the Howard Road Bridge [Brooks 19831. The web penetration in 12 years France corresponds to an average penetration of 16 mils (400 m) per year and per surface. This rate is about the same as the highest The French specification for weathering steels cautions the measured in Michigan. It may be an upper bound under adverse user that the formation of the protective oxide coating depends conditions of exposure. not only on the chemical composition but also on environmental 28 factors (humidity, atmospheric pollution, and wet-dry cycles), determine in each case if the protective coating will form under overall and detail design, and exposure conditions [AFNOR the specific exposure conditions [DDR 1973]. 1979]. In applying weathering steel to structures, the user is Weathering steels can be used without restrictions if the cor- directed to take all necessary precautions that will ensure the rosion penetration, as a function of time of exposure, does not existence of conditions favorable to the formation of a protective exceed specified maximum values (Fig. 47). Guidelines for their coating. These measures must prevent ponding of water, trap- application in rural, urban, and industrial environments are ping of humidity, and permanent humid conditions fostered by given as a function of the sulfur dioxide content of the atmo- condensation or capillary action. Rain water must be able to sphere, and the degree of aeration and sheltering of the structure run off, and humidity must be able to dry without hindrance. [DDR 1983]. Weathering steels may be used without corrosion Particular attention should be given to low points of support protection in a 1.25-mile (2-km) coastal zone along the North and sheltered areas. Sea provided that even stricter limits on sulfur dioxide contents are met and that the steel is not exposed to salt water. The expected loss in section due to corrosion must be considered in Federal Republic of Germany determining the safe load-carrying capacity of a structure. The guidelines also point out that the structural details in- The application of bare weathering steel in highway bridges fluence the aeration of the structure, time of wetness, and cor- on the federal system is only permitted with special approval rosion performance. of the Department of Transportation. This requirement virtually amounts to a ban. Since its inception in 1980, only two excep- JAPANESE EXPERIENCE tions were granted, namely for composite girder bridges over electrified railways. This was done to avoid having to restrict Atmospheric corrosion resisting steels, first introduced in Ja- railway traffic during maintenance painting, had the bridge been pan in the 1960's, are increasingly being used for highway and built of regular steel [DASt 1979, FRG 1982]. railroad bridges. The initial motivation was mainly to prolong Weathering steel bridges built before 1980 are exhibiting large the life expectancy of paint. Numerous atmospheric exposure differences in corrosion performance. The relationship between tests of painted steel plates with scratches had indicated a sig- macroclimate (rural, urban, industrial, and marine) and cor- nificantly lower corrosion along the scribe marks when the rosion rate is found to be difficult to assess quantitatively. Poorly underlying steel was of the corrosion resisting type. The prospect aerated and humid parts of the steel structure are performing of longer paint life encouraged the use of painted weathering very unsatisfactorily [FRG 1982]. The restrictions on the use steels in relatively high humidity environments and in industrial of weathering steel were imposed because the corrosion rate of areas. existing structures is not declining as anticipated. The steel Because repainting costs were increasing, bridge owners periodically sheds the rust layer, depending on the atmospheric started, albeit cautiously, to use weathering steel in an unpainted conditions at the bridge site. condition in the 1970's. Many of these structures were treated The use of weathering steel has never been permitted for once with a so-called "weather coat," a phosphorus-rich paint railway bridges because of concerns about its fatigue perform- like surface coating. This sacrificial initial coating is believed to ance. For the same reason, exceptions are not granted for high- help the steel form a stable and protective oxide. It also has the way bridges with orthotropic weathering steel decks. advantage of hiding the early stage of rusting and preventing The weathering steel specifications set limits on atmospheric large amounts of iron oxide in the runoff water from staining pollution [DASt 1979]. If these are exceeded, the steel must be concrete piers and abutments. Japanese designers consider the painted. Corrosion protection is called for when the time of aesthetics of a bridge to be important, particularly when it is wetness is long or the steel is subjected to chloride contami- built in a heavily populated area. nation. According to a survey conducted by the Japanese Society of For purposes of designing load-carrying members, the thick- Civil Engineers, 115 weathering steel bridges had been built in ness must be increased to account for net section loss. This Japan by March 1979. Of those, only a few were of nonpainted allowance varies with service life and environment. It is at most steel. /6 in. (1.5 mm) per side for structures in industrial and marine Since then, Japanese engineers examined numerous aged environments with a 60-year service life (Fig. 47). bridges, unpainted bridges, both domestic and foreign, and The corrosion performance of existing structures must be buildings to understand the possible problems of building with monitored with specimen prepared and installed at the site in unpainted weathering steel. Convinced of its feasibility, bridge accordance with set guidelines [DASt 1979]. owners began to construct in the late 1970's and early 1980's unpainted highway and railroad bridges, but they carefully designed details in a manner that would help the steel form a German Democratic Republic. protective oxide coating. These bridges are periodically monitored to obtain more information on their long-term behavior. The conditions for the formation of a protective oxide coating Because weathering steels were first used for painted steel are said to exist in the macroclimate of the German Democratic bridges, the chemical composition of the steel was mainly in- Republic, provided that: (1) the steel surface does not remain tended to provide better weldability rather than better weath- humid for extended periods of time; and (2) the atmospheres ering characteristics. Recent application to nonpainted can be classified as rural and urban, as defined in Standard TGL construction led to the inclusion of two separate chemical com- 18704. For exposures in severe industrial and marine atmo- positions in the proposed modification of JIS G 3114, one for spheres and in other corrosive environments, the user must painted and one for nonpainted weathering steels. 29

Table 13. Nonpainted weathering steel highway bridges in Japan.

Weight Type of Bridge Name Site [Ton] Structure Environment Completion Name

Muranaka-Kobashi Br. Komaki, Nagoya Simple spam - 1967 Painted in 1982

Kawatetsu Chita in the Chita Works 60 Simple supported Sea-shore, 1967 Now, 1300m irom sea- No. 2 Br. of Kawatetsu Corp. composite girders shore (8-years, 200m)

Daiichi Ryogoku Br in the Fukuyaiia Steel 860 Simple supported Sea-shore, 1969 - Works of NIPPON Kokan composite girders partially on and others the sea

Kurogane Br. in the Hirohata Works 120 Simple supported Sea-shore 1973 - of NIPPON Steel Corp. composite girders

Shin-Ohashi B., Tokyo 662 Simply supported on the sea 1974 Removed now Temporary girders

Sanjin Br. Hyogo 92 Simple supported Rural 1978 - curved box girders

Mukogawa Br. Hyogo 110 Simple supported Rural 1980 - composite girders

Dejima Off-Ramp Br. Osaka 120 3 spaned continu- Industries 1980 Tentatively erected in ous un-composite the sea-shore for girders 2 years

Kokubgawa Br. Kochi 291 Simply supported Rural 1981 - composite girders

So far, only a few nonpainted weathering steel highway tered interior girders, however, exhibited a somewhat coarse bridges have been built in Japan, as listed in Table 13. Some, and flaky oxide coating. Bridge engineers, who inspected the in private use, are monitored periodically. The Kawatetsu Chita bridge over the years, felt that the concept of nonpainted weath- No. 2, Daiichi Ryogoku, and Kurogane bridges, located on the ering steel construction was feasible provided that rain water is grounds of steel companies, have painted, weather coated, or drained, debris is cleaned up or prevented from accumulating, nonpainted sections. This allows the companies to periodically and splash water does not reach the steel. monitor and compare the performance of the three systems. It The 1,547-ft (472-m) long Daiichi Ryogoku-bashi Bridge was should be noted that these early experimental bridges were con- built in 1969 in the Nippon Kokan Company's (NKK) Fuku- structed near the seashore, where the environmental conditions yama Works, Hiroshima Prefecture. It crosses a channel, and are not favorable for weathering steels. On the other hand, if a part is over the sea. The steel is Cupten 50. Two simple-span all coastal areas were excluded, there would not be many sites girder sections near the ends were left nonpainted and are in- left on which to build weathering steel bridges in Japan. The spected periodically. In 1979, 10 years after construction, the information obtained from these bridges is, therefore, considered nonpainted sections were generally covered with a stable rust to be extremely important. The performance of the four bridges coating. The outsides of the exterior girders had a particularly listed at the top of Table 13 is summarized in the following. good appearance typical of weathering steels. However, the ex- The Muranaka-Kobashi Bridge is believed to the first un- pansion joints, supports, and details near the supports exhibited painted weathering steel bridge built in Japan for public use. rapid and flaky corrosion. Excessive corrosion pits were also The bridge is located in Komaki, a rural area near Nagoya. found. This unfavorable corrosion was believed to be caused by When inspected in 1976, 9 years after construction, the bridge runoff water leaking through the expansion joints in the bridge was found to have a fairly good protective oxide coating on the deck and keeping the steel humid. It was, therefore, recom- webs and on most flanges [Yamada 1983a]. However, some top mended that the structural details near the end of the bridge surfaces of the tension flange were flaking. Since these places needed to be modified to prevent the runoff water from reaching were not washed by rain, the accumulated rust and debris had the steel structure. trapped moisture and kept the steel humid for extended periods The Kurogane Bridge was erected in 1973 in the Hirohata of time. The bridge was painted in 1982. Works of Nippon Steel Co., near Himeji, Hyogo Prefecture. The Kawatetsu Chita No. 2 Bridge was erected in 1967 in This bridge is also privately owned. Sections were weather the Chita Works of Kawasaki Steel Co., near Nagoya. The site coated or left nonpainted to evaluate the long time performance is located on reclaimed land, first about 660 ft (200 m) and of weathering steel under these conditions. The bridge consists later 4,260 ft (1300 m) from the sea because of further recla- of eight simple-span composite girders made of A588 Grade A mation. The girders are divided into five sections that were left rolled sections. It is located near the seashore. A survey of bridge unpainted, weather-coated, and painted with three different engineers, who inspected the bridge in 1979, showed that most types of paints. The exterior girders, running north—south, are of them agreed that the exterior girders of the nonpainted spans dried by the morning and afternoon sun. They formed in general had formed a stable oxide coating typical of weathering steels. a protective oxide coating typical of weathering steel. The shel- However, about one-half of the engineers were not convinced 30 that corrosion of the interior girders, laterals, bracings, bearings, over weather coating, as far as appearance was concerned. and shoes had stabilized. The nonpainted steel surfaces exhibited The case studies of the three aforementioned bridges owned a nonuniform appearance because of differences in corrosion by Japanese steel companies and the deliberately cautious ap- rates and water running over the steel, but 70 percent of the proach of Japanese bridge engineers convey a desire to make bridge engineers considered this to be negligible. On the other nonpainted weathering steel bridges perform as intended the hand, the nonuniform appearance of the weather-coated surfaces first time around. Chapter Eleven gives examples of bridge de- led 80 percent of the engineers to favor nonpainted construction tails the Japanese use to help achieve this goal.

CHAPTER THREE

CORROSION MECHANISMS

BEHAVIOR OF STEEL IN THE ATMOSPHERE As to the currently considered mechanisms for explaining the atmospheric corrosion process, three will be described. The deterioration or corrosion of steel may simply be described as a consequence of exposure to the atmosphere. In a more precise sense, it is the reaction in air under moist conditions aided air pollutants, principally the sulfur oxides. The chemistry Humidity Film Reactions of the reaction is defined as being electrochemical in nature, because it involves ionization and transfer of electrons, as do When iron is attacked under humid conditions in a clean most chemical reactions. The most frequently used metals (iron, rural atmosphere, the first oxidation product is ferrous ion in aluminum, and zinc) are found in nature in the form of ore. A the lowest oxidation state, namely, Fe". Because of the presence great deal of energy must be expended to release or reduce the of air dissolved in moisture, the ferrous ion can react with it ore to the metallic state. Obviously, these metals have a great and precipitate ferrous hydroxide, which can be quickly oxidized deal of pent up energy to return to the mineral state. Aluminum further to the ferric state, Fe", to give the gelatinous precipitate and zinc react in the air to form thin protective films of alu- of ferric hydroxide. This series of reactions can be expressed in minum oxide and basic zinc carbonate, respectively. Iron or a fashion suggested by Kunze and others as follows [Kunze steel does likewise, but the oxide film is unfortunately nonpro- 1974]: tective. At the present time the most interesting and least completely The anodic iron dissolves in the condensed moisture film. understood behavior is the controlling mechanism of the high- strength low-alloy steels, beginning with the simplest copper- Humidity ±2 bearing steels that contain as little as 0.04 percent copper. Here, Fe Fe + 2e (1) for the first time, it has been observed that a self-limiting or stifling of corrosion is occurring in a mild carbon steel. The To counter balance this reaction, the cathode accepts the early trial and error compositions of Buck, Taylerson, and Lar- electrons and passes them on to oxygen which is converted to rabee, and the work by Larrabee and Coburn, have resulted in hydroxyl ion. the ultimate development of the currently available proprietary weathering steels. The period of the late 1960's and early 1970's 02 + 2H20 + 4e - 4 OH- (2) marked the beginning of a worldwide research interest into the nature of the corrosion products of various weathering steel With free ferrous ions and free hydroxyl ions in solution, compositions. Progress was expedited and theories evolved by the following reaction occurs: virtue of the availability of much improved analytical instru- mentation. Fe 2 + 2 OH- - Fe(OH)2 (3) The familiar iron compounds are: Fe(OH)2, ferrous hydroxide; Fe(OH)3, ferric hydroxide; FeO,, iron oxide or rust; and The fresh ferrous hydroxide is then oxidized by air to Fe304, magnetite. However, numerous other species of iron com- produce hydrated ferric oxide. pounds have been found and the simple representations are inadequate to explain what is already known. The most recent Rust species isolated are the ferric oxyhydroxides known as a-FeOOH, Fe (OH)2 + 1/2 02_ Fe OOH + H2O (4) $-FeOOH, 'y-FeOOH, and 8-FeOOH. They have been examined by X-ray diffraction, infrared and far infrared. Each is different The following reactions may take place instead of those listed from the other. Knowledge of their existence is only about 20 in steps 2 and 3. The final product in both cases is hydrated years old. ferric oxide. 31

Fe 3 + 3 OH- - Fe(OH)3 (5) 1964, 1967]. In addition, despite the reduction in supply of SO2, the corrosion can continue on an electrochemical basis instead 2 Fe(OH)3.Deh)drat2_2 FeOOH + 2 H2O (6) of by acid attack [Evans 1972a]. Kunze [1974] confirmed this cycle of events. 4. When the supply of oxygen is not sufficient, the reaction takes place according to Eq. 7 instead of Eq. 4:

3 Fe(OH)2 + 1/2 02 Fe364 + 3 H20 (7) Electrochemlcal Cycle Several investigators proposed a two-layer model of the oxide Or, as suggested by Huller [1966], according to Eq. 8. during the advanced stable stages of rusting [Horton 1965, Evans 1972a, Bohni 1979, Misawa 1971-1974, Kunze 1974]. Photo- Fe(OH)2 + 2 FeOOH - Fe304 + 2 H2O (8) micrographs of rusts on low-alloy steels support the idea of a two-layer oxide coating [Horton 1965]. The aforementioned humidity film reactions are rather slow. Kunze described in detail a two-layer model. As illustrated Moist air, even if saturated, produces little rusting of iron in in Figure 1, the model consists of an inner layer bounded by the absence of dust and pollutants [Evans 1972a]. Hence, one the base metal and the oxidation-reduction front (red-ox front) has to consider more active stimulants, such as the pollutants inside the rust coating, and an outer layer bounded by the red- in an industrial environment, to explain the more rapid corrosion ox front and the atmosphere. Kunze based his model on the of iron observed in service. following findings. First, during the advanced, stable rusting stage, no cathodic diffusion current was measured in the outer layer. This indicates that electrochemical reactions, as explained later, cannot occur in the outer oxide layer. Second, the cathodic Acid Regeneration Cycle current density in the inner layer increases as the thickness of the rust layer increases. In other words, as the rust thickens, it The most common pollutants in an industrial environment provides a barrier to further corrosion, in effect slowing down are the sulfur oxides produced during the combustion of fossil the rate of corrosion attack. Third, the curve of current density fuels, such as oil and coal, each containing several percent of versus potential in the inner layer is the same for an electrolyte combined sulfur. Upon oxidation, sulfur dioxide is the predom- free of oxygen. This indicates that the electrochemical reactions inant gas with the remainder, about 10 percent, being sulfur which are taking place in the inner layer do not directly involve trioxide. The conversion of these gases to the mixed sulfurous oxygen. Oxygen is consumed only in the outer layer and at the and sulfuric acids becomes the source of the corrosive pollutant. interface with the base metal. Furthermore, because the cor- This cycle, mentioned in Vernon's early work between 1943 and rosion process does not cease, either in water solutions or during 1949, is called the acid regeneration cycle in the literature. It natural and artificial weathering, and because the hydrated ferric is a second mechanism for explaining the atmospheric corrosion oxide content of the oxide coating is not used up despite constant of steels and consists of the following reactions: reduction, one must assume that the oxygen in the air reoxidizes the magnetite (FeO,) in the outer layer to hydrated ferric oxide. Sulfuric acid forms in the presence of rust, with rust acting However, the reduction cannot take place electrochemically, as the catalyst. because there is no cathodic diffusion current in the outer layer. It must take place by chemical reactions. Rust The conclusion is that the third mechanism of rusting involves SO2 + HO + 1/2 02- H2SO4 (9) an electrochemical cycle. The details are as follows:

The generated sulfuric acid chemically attacks the iron to Some rust forms on the steel surface during the acid re- form ferrous sulfate. generation cycle, by cathodic oxygen reduction. This 'step was already described. 2 H2SO4 + 2Fe + 02 -, 2 H20 + 2 FeSO4 (10) An electrochemical cell is established. It involves oxidation, reduction, an anode (corroding metal), a cathode (existing Ferrous sulfate is oxidized to ferric sulfate and hydrated rust), and an electrical current through the electrolyte (water ferric oxide (rust). solution). At the metal-to-rust interface, the anodic region, iron lib- 6 FeSO4 + H2O + 3/2 02 - 2 Fe2(SO4)3 + 2 FeOOH (11) erates electrons

And, finally, ferric sulfate hydrolyses to additional rust. Fe - Fe 2 + 2e (13)

Fe2(SO4)3 + 4 H20 - 2 FeOOH + 3 H2S0.I (12) and, in the presence of sulfates in the rust, -ferrous sulfate forms.

Equation 12 indicates that all consumed acid is theoretically Fe 2 + SO2 - FeSO4 (14) regenerated. Actually, this acid regeneration cycle is limited, as it has been reported that each molecule of SO2 can lead to the The electrons flow to the cathodic region near the red-ox front formation of from 15 to 40 molecules of rust [Schikkor 1963, via the magnetite. As Schwarz [1972] pointed out, other iron 32 compounds containing Fe" and Fe 3 can also transmit the enough oxygen from the air penetrates the outer layer and electrons. reaches the red-ox front. Formation of a protective rust in the In the inner layer near the red-ox front, the cathodic re- outer layer inhibits continuous supply of oxygen and water to gion, rust is reduced to magnetite by either of the following this front. As a result, the corrosion attack may slow down reactions: considerably.

8 FeOOH + Fe' + 2e - 3 Fe0 + 4 H20 (15) In summary, the red-ox front separates the rust coating into an inner layer, where rust is electrochemically reduced, and an or outer layer, where magnetite is chemically oxidized to rust. The red-ox front is not stationary. It moves, to some degree, inwards 6 FeOOH + 2 H + 2e - 2 Fe304 + 4 H20 (16) and outwards depending on the humidity level and the oxygen supply. The first of the two reactions requires Fe ions and the The previously described corrosion mechanisms apply to second H ions. The latter come from hydrolysis of ferrous steels in general. This is because marked differences between sulfate in the anodic region of the inner layer. the atmospheric corrosion behavior of plain carbon steel and weathering steel do not start to show up until a fully developed rust-layer has formed [Horton 1965]. The reasons why weath- 2 FeSO4 + 3 H20 + 1/2 02 - 2 FeOOH ering steel develops a protective rust layer are discussed• in the +4H+2SO4 (17) next section.

The magnetite that forms in the cathodic region reoxidizes again to rust. WEATHERING STEEL

3 Fe30 + H20 + 0 - 9 FeOOH (18) Structural carbon steels become atmospheric corrosion resistant when they are modified by the inclusion of 2 percent or less of such common alloying elements as copper, phosphorus, or chromium, nickel, and silicon. To explain this behavior, the corrosion mechanisms of weathering steel reported in the lit- 2 Fe304 + H20 + 0 - 6 FeOOH (19) erature will be reviewed. To begin with, one must accept that corrosion science has proven beyond doubt that the corrosion of metals, once initiated, proceeds by an electrochemical mech- These chemical reactions take place at the red-ox front, when anism.

Atmosphere $04 02 H2O

Chemical Oxidation Outer Oxide 3Fe3O4+ 92H 20+ O2— 9FeOOH Layer FeOOH or: 20+ 6FeOOH 2Fe304+ 3H O2—_- -. Red-ox front ------+consumed H+ Cathode: Electrochemical Rust Reduction 8FeOOH + Fe 2+2e-3Fe3O4+4H2O Inner or: Oxide 6feOOH + 2H +2e-2Fe304+4H20 Layer

Fe3O4 Fe Hydrolysis:

FeSO4+-+20+-02---*-FeOOF{+H2SO4 H produced bV hydrosis

Anode: Steel Fe _____-> _F e +2 +2e - and Fe 2+ SO4 2 —.FeSO4 in presence of SO2

Figure 1. Schematic representation of two-layer oxide modeL [Kunze 19747 33

The existence of anodic and cathodic regions in the inner rust Fe(OH)2 - FeOOH + FI2O (20) layer implies that cations diffuse outwards through the pores of the inner rust layer, which are filled with electrolyte when the In other words, each of the different corrosion products, 3 _ 2 found on the surface steel is wet. Any means by which the pores are plugged retards y-FeOOH, a-FeOOH, and FeO(OH) 1 the corrosion process. In all steels, the cations consist mainly of naturally weathered steel specimens is basically one form of partially dehydrated ferric hydroxide. The supply of oxygen and of diffusible H ions and Fe 2 ions. In weathering steels that water, the pH value, and the presence of alloying elements are alloyed with copper, chromium, and nickel, the Cu", Cr 3, and Ni" ions strongly tend to form compounds that are left govern the degree of dehydration and the crystal system. behind because of their large diameter and small diffusion ve- The a-FeOOH mentioned above is obtained from solid state locity. For this reason, Cu", Cr", and Ni" concentrations are transformation of the amorphous ferric hydroxide when the rust found in the inner regions of the rust coating of weathering dries and the amorphous ferric oxyhydroxide loses water. steels containing these alloying elements [Bruno 1973, Kunze Solid State Transformation 5 a-FeOOH 1974]. 5FeO(OH)32. As was described earlier, the red-ox front separates the rust + 3 H20 (21) coating into an inner layer and an outer layer. Changing supplies of moisture and oxygen move the red-ox front inwards and Misawa reported remarkable differences in the rust coating outwards, as the steel undergoes wet-dry cycles during natural that formed on plain carbon and low-alloy steels during natural weathering. weathering. The rust on low-alloy steel exhibited uniform thick- During the wetting part of the cycle, the red-ox front moves ness, and the rust-to-steel interface was even. In contrast, the outwards. Diffusion of cations increases the supply of hydrogen rust on plain carbon steel was not uniform and the rust-to-steel ions as well as copper, chromium, and nickel cations. The en- interface was uneven. Misawa attributed the uniformity to the vironment within the electrolyte becomes acidic and the alloying presence of alloying elements. Because copper, phosphorus, and elements, cannot precipitate. chromium are enriched in the inner rust layer of low-alloy steel, During the drying part of the cycle, the red-ox front moves the dissolution of steel, the formation of y-FeOOH, and sub- inwards. It loses hydrogen ions, the pH value increases, and sequent precipitation of the amorphous compound would also compounds rich in copper, chromium, and nickel are precipi- be uniform. Moreover, the amorphous ferric oxyhydroxide rust tated. For example, the copper ions in the electrolyte precipitates on low-alloy steels contained a considerable amount of bound as basic copper sulfate once a critical concentration of ions is water, as compared with that on carbon steel. Hence, one can reached after repeated wet-dry cycles. This process gradually assume that the rust on low-alloy steel dries more slowly. The plugs the pores and makes the layer denser, thus inhibiting the rate of solid state transformation of the amorphous ferric ox- outward diffusion of Fe 2 ions and the penetration of moisture yhydroxide, as shown by Eq. 21, is slower for low-alloy steel and oxygen. Hence, both the anodic and the cathodic reactions than for carbon steel. Horton's [1965] work had also indicated slow down, and the rate of corrosion diminishes. Other inves- that the rust on low-alloy steels dried out more slowly than on tigators have been unable to confirm the presence of basic copper carbon steel. sulfate [Suzuki 1980]. The uniformity of the rust and the slower rate of drying, both Additional theories about the formation of the protective ox- induced by the presence of alloying elements, help the low-alloy ide film on weathering steel have been suggested. Misawa et al. steel develop a compact rust coating with fewer cracks. Plain analyzed the rust coating with a wide variety of advanced an- carbon steel, on the other hand, has a nonuniform, porous rust alytical instruments [Misawa 1971, 1973, 1974a, 1974b]. They coating with many cracks that let water, oxygen, and other identified the chemical compounds whose physical features can contaminants freely penetrate. Figures 2 and 3 show a schematic account for the beneficial performance of copper-bearing steel representation of the oxide coating on carbon steel and low- and more particularly the weathering steels. Their work iden- alloy steel, respectively. tifies the role of copper and phosphorus. Misawa et al. concluded that rusting initiates by the aerial oxidation of ferrous ions, which are dissolved from the steel because of a slightly acidic thin water layer on the steel surface. For example, fresh rain dissolves atmospheric pollutants, mainly sulphur dioxide, and becomes acidic with a pH value of four. The low pH water layer covering the steel surface leads to the formation and precipitation of y-FeOOH. The Greek alphabet prefix, y, specifies the crystalline structure of the rust formed in this process. The 7-FeOOH dissolves in the slightly acidic rain water during the wetting periods and precipitates as amorphous ferric oxyhydroxide, FeO(OH)3_, upon drying. Analysis of the rust indicated that x = 0.4 for the latter compound. This suggests that the chemical formula for the amorphous matter can be written as Fe5013H11. One can assume that Fe5O13H11 is composed to two molecules of hydrated ferric oxide and three molecules of ferric hydroxide. The hydrated ferric oxide is a dehydrated form of ferric hydroxide. The dehydration reaction Figure 2. Schematic representation of the oxide coating on weath- is as follows: ered plain carbon steel. 34

In 1980, Suzuki et al. examined natural and artificial rust layers that formed on a series of steels, from pure iron to weathering steel [Suzuki 1980]. They concluded that: (1) copper ad- Electrolite ditions to iron accelerate the formation of amorphous substances and increase the mechanical strength of the rust coating; and -,'-FeO0H FeO(OH)32 (2) beneficial alloying elements decrease the formation of inter- - 4So2 '°2 ,H20 mediate substances, which would be converted by cathodic re- i-u-u- 4J 4LLUJ I Lfl ((1NJU action to crystalline magnetite (Fe3O4) and, thereby, decrease - _ the electrical conductivity of the rust coating. H Ill EFFECT OF AQUEOUS ENVIRONMENTS AND SALT Steel/Rust Interface1 The formation of a protective rust film on weathering steel requires alternate wet-dry cycles of natural weathering and a Figure 3. Schematic representation of the oxide coating on weath- salt-free atmosphere. Under prolonged periods of wetness and ered low alloy steel. under severe exposure to deicing salts, weathering steel corrodes at the same rate as plain carbon steel [Larrabee 1958, Schmitt 1967, Zoccola 1976, Tinklenberg 1980, Reed 1982, Brocken- brough 1983]. If weathering steel remains wet for a long period of time, the In summary, there are two plausible mechanisms that give necessary conditions are not met for the formation of the amor- weathering steel its corrosion resistance. One suggests that the phous ferric oxyhydroxide, suggested by Misawa et al. [1974b], corrosion resistance is achieved by precipitation of basic com- or the periodic inward/outward movement of the red-ox front pounds rich in alloying elements. The other presumes that the resulting in precipitation of basic compounds rich in alloying alloying elements catalyze the precipitation of amorphous ferric elements, as described by Kunze [1974]. oxyhydroxide. The two mechanisms have several common fea- The deleterious effect of deicing salt on atmospheric corrosion tures: two-layer structure of the rust coating, electrochemical resistance of steels results from several phenomena. First, deicing corrosion cycle, enrichment of alloying elements in the inner salt dissolved in runoff water creates an aggressive electrolyte. rust layer, and basicity of the rust. The differences may be more When the negatively charged chloride ions, Cl-, come in contact a matter of viewpoint than substance. with the steel, they shift the already negative potential of the For reference purposes, some other explanations are also listed steel to larger negative values. The ensuing increase in the gal- in the following. An early attempt to explain the improved vanic potential accelerates metal corrosion [Uhlig 1973]. Other atmospheric performance of copper-bearing steel over that of negatively charged ions, such as 5o4_2 and NO 3 which originate carbon steel was made by Carius. He proposed that the resistance from industrial pollutants, have a similar though less pro- to atmospheric corrosion increased when a metallic copper film, nounced effect on steel. or a copper-rich layer, form next to the steel [Carius 1930]. Secondly, when chloride ions are present, the rust coating However, as Copson [1945] pointed out later, this film of copper contains large amounts of /3-FeOOH. This form of hydrated has not been observed experimentally, nor have artifically ap- ferric oxide is crystalline. In contrast to y-FeOOH, it does not plied films of copper been as effective in reducing corrosion as convert to the protective amorphous ferric oxyhydroxide. directly alloying the copper in the steel. Thirdly, the strong tendency of salts to absorb moisture from In 1945, Copson reported finding higher sulfate levels in the the atmosphere lowers the critical relative humidity at which rust coating of copper bearing steels than in mild carbon steels steels begin to significantly corrode. In other words, the presence [Copson 1945]. He attributed the beneficial influence of copper of salt raises the relative humidity in the immediate neighbor- and nickel to the formation of basic copper and nickel sulfates, hood of the steel, lengthens the wetting periods, and hinders respectively. Vernon [1960] later characterized the basic copper the drying process. The chloride ions are reused; they do not sulfate as [Cu(OH)2Cu,,] SO4. These basic sulfates were believed need to be resupplied. to plug the pores in the rust and to render it less permeable to For the foregoing reasons, the superior corrosion resistance oxygen and water. exhibited by the weathering steels, when properly utilized, is In 1970, Fyfe et al. examined the corrosion behavior of eight not usually observed in aqueous and saline environments. There- pure iron alloys, one mild carbon steel, one copper-bearing steel, fore, under less than boldly exposed conditions, weathering steels and one A588 Grade A steel exposed to rural, industrial, and cannot necessarily be assumed to have better corrosion resistance marine atmospheres in England [Fyfe 1970]. They stated that than carbon steel. the primary purpose of alloying with copper was to nullify the deleterious effect of sulfur, which can come from two sources: (1) present in the steel as ferrous sulfide, FeS, or manganese EFFECT OF PRIMARY ALLOYING ELEMENTS sulfide, MnS; and (2) introduced from the atmosphere either FROM LARRABEE-COBURN INVESTIGATION directly as H2S or from the cathodic reduction of SO2. The mechanism is somewhat speculative, because it is not clear how Larrabee and Coburn [1961] performed the most extensive even small amounts of copper (as low as 0.05 percent) could and systematic study of steel compositions alloyed with the nullify all of the available sulfur ions. elements copper, phosphorus, chromium, nickel, and silicon that 35 are known to contribute most to the atmospheric corrosion Table 14. Identification of curves plotted in Figures 4 through 18 and resistance of the weathering steels. They arranged 270 steels in chemical composition of steels. [Larrabee 1961] a five-way factorial experiment with five levels of copper content Curve Steel Identification No.6 Composition, Percent (0.01 percent mm. to 0.51 percent max.), three of phosphorus No. Cu P Cr Ni Si (0.01 percent mm. to 0.12 percent max.), three of chromium (0.1 percent mm. to 1.3 percent max.), two of nickel (0.05 (a) Effect of copper (Figs. 4 to 6) percent mm. to 1.1 percent max.), and three of silicon (0.1 1 1 2 3 4 5 - - - - 2 6 7 8 9 10 - - 1.0 - percent mm. to 0.64 percent max.). Of the nine grades of A588 3 11 12 13 14 15 - 0.64 - - 4 21 22 23 24 25 - 1.30 - - steel, the following have chemical requirements for these ele- 5 31 32 33 34 35 - - - 0.20 6 61 62 63 64 65 - - - 0.52 ments that fall entirely within the range studied by Larrabee 7 91 92 93 94 95 0.06 - - - 8 131 132 133 134 135 0.06 0.62 - 0.22 and Coburn: for copper, seven grades; for phosphorus and chro- 9 136 137 138 139 140 0.06 0.62 1.0 0.18 10 181 182 183 184 185 0.10 - - - mium, all nine grades; for nickel, eight grades; and for silicon, 11 261 262 263 264 265 0.10 1.20 - 0.58 seven grades. Obviously, the Larrabee-Coburn data are the key 12 266 267 268 269 270 0.10 1.20 1.0 0.53 to understanding the role of these five elements. All 270 steels (b) Effect of phosphorus (Figs. 7 to 9) had less than 0.1 percent carbon, 0.25 percent to 0.40 percent 1 1 91 181 - - - - manganese, and less than 0.02 percent sulphur. Replicate cou- 2 4 94 184 0.21 - - - 3 5 95 185 0.47 - - - pons were mounted on exposure racks at a 30-deg angle from 4 6 96 185 - - 1.0 - 5 11 101 191 - 0.66 - - the horizontal facing south. They were exposed for 15.5 years 6 21 111 201 - 1.30 - - 7 31 121 211 - - - 0.24 at semirural South Bend, Pennsylvania, industrial Kearny, New 8 41 131 221 - 0.62 - 0.22 9 45 135 225 0.44 0.65 - 0.21 Jersey, and marine Kure Beach, North Carolina, at a distance 10 46 136 226 - 0.64 1.0 0.24 11 50 140 230 0.45 0.64 1.0 0.18 of 800 ft (240 m) from the shore. 12 61 151 241 - - - 0.55 The data are examined hereafter for the effect of each alloying 13 81 171 261 - 1.30 - 0.59 14 85 175 265 0.44 • 1.20 - 0.52 element alone, in combination with a second element, and in 15 86 176 266 - 1.20 1.0 0.49 16 90 180 270 0.45 1.30 1.0 0.50 combination with the remaining elements. Figures 4 through 18 (five elements times three environments) show the average cor- (c) Effect of chro,nijn (Figs. 10 to 02) rosion penetration per side as a function of the element con- 1 1 11 21 - - - 2 4 14 24 0.23 - - - centration and the environment. Each data point represents one 3 5 15 25 0.46 - - - 4 6 16 26 - - 1.0 - of the 270 steels. The curves connect the data points for the 5 31 41 51 - - - 0.27 6 61 71 81 - - - 0.59 steels with nearly equal concentrations of four of the five alloying 7 91 101 111 - 0.06 - - elements. The concentrations of the fifth element vary along the 8 121 131 141 - 0.06 - 0.27 9 125 135 145 0.45 0.06 - 0.19 abscissa. The chemical composition of the steels, corresponding 10 126 136 146 - 0.06 1.0 0.23 11 130 140 150 0.45 0.06 1.0 0.17 to the number on the end of each curve, are identified in Table 12 181 191 201 - 0.09 - - 13 241 251 261 - 0.11 - 0.55 14. The curves were cut off at 16-mils (400-tm) penetration. 14 245 255 265 0.46 0.10 - 0.57 15 246 256 266 - 0.11 1.0 0.49 Some steels in Figures 5 and 7 are identified with an underlined 16 250 260 270 0.45 0.10 1.0 0.60 individual composition number, written next to the corresponding data points. They correspond to the Larrabee-Coburn com- (d) Effect of nickel (Figs. 13 to 15) position numbers given in Table 14. 1 1 6 0.045 - - - 2 2 7 . 0.045 - - - For ease of comparing the A588 specification requirements 3 4 9 0.22 - - - 4 5 10 0.46 - - - for chemical composition with the Larrabee-Coburn data, the 5 11 16 - - 0.61 - 6 21 26 - - 1.30 - concentration ranges of copper, phosphorus, chromium, nickel, 7 31 36 - - - 0.23 8 61 66 - - - 0.59 and silicon for each grade are shown at the top of Figures 4, 9 91 96 - 0.06 - - 10 131 136 - 0.065 0.61 0.24 7, 10, 13, and 16. 11 134 139 0.20 0.06 0.60 0.17 Table 14 identifies for each curve the steel composition num- 12 135 140 0.45 0.055 0.65 0.20 13 181 186 - 0.09 - 0.55 bers assigned by Larrabee-Coburn and the average concentration 14 261 266 - 0.115 1.20 0.53 15 264 269 0.23 0.095 1.25 0.53 of the alloying elements. Table 15 compares the "reduction in 16 265 270 0.45 0.10 1.25 0.60 thickness per side" or the "corrosion penetration" of the steels containing one or two alloying elements. (e) Effect of silicon (Figs. 16 to 18) 1 1 31 61 - - - - 2 4 34 64 0.23 - - - 3 5 35 65 0.48 - - - 4 6 36 66 - - - 1.0 Copper 5 11 41. 71 - - 0.64 - 6 2151 81 - - 1.30 - 7 91 121 151 - 0.06 - - Figures 4, 5, and 6, respectively, show the effect of copper 8 101 131 161 - 0.06 0.61 - 9 105 135 165 0.45 0.06 0.61 - content on the corrosion penetration in the rural, industrial, 10 106 136 166 - 0.07 0.57 1.0 11 110 140 170 0.45 0.06 0.66 1.0 and marine environments. The curves 1 are for steels alloyed 12 181 . 211 241 - 0.10 - - 13 201 231 261 - 0.10 1.20 - with copper alone. Their corrosion penetration in the three 14 205 235 265 0.46 0.09 1.30 - 15 206 236 266 - 0.11 1.20 1.0 environments is given in the top five lines of Table 15(a). The 16 210 240 270 0.44 0.08 1.30 1.0 left most points on these curves correspond to carbon steel (0.02 percent Cu max.) which exhibited a corrosion penetration of a. The steel Identification nombers are the some as in Ref. (Larrabee 1961J. 12.3, 28.8, and 52.0 mils (312, 732, and 1,320 p.m) in the rural, industrial, and marine environments, respectively. Moving along curves 1 to the right, it is evident that the initial 0.05 percent copper gave the greatest reduction in corrosion penetration in

COPPER CONTENT IN GRADES OF A588 STEEL 4O; A KEARNY, N.J B 1 Carbc C 5.5 Years D 350 E ).% F '.00% H J -C - -K 300 300

250- 250 SOUTH BEND. PA. E 5.5 '(son z 200 200

CO, steel, IL COPPER STEELS ISO 0.52sf 5 0.52 Si 6 1.0 Ni 100 8,4,2,7 . NI I-OP 100 1.3 Cr 10 GS p 10 1 3Cr 94 0-lOP 1.20 Cr, 0.58 Si QIO J.2O Cr, O.58SI 50- 50 O 1.20Cr, l.ONi,0-53Si 1 II 12 11 12 265 0-lOP, 1.20Cr, ONi, 0.53 Si

0.1 02 03 0-4 0-5 06 0 06I 0-2 0.3 0.4 05 0- COPPER CONTENT, S COPPER CONTENT, %

Figure 4. Effect of copper content on corrosion penetration at Figure 5. Effect of copper content on corrosion penetration at South Bend, Pa. rural site, 15.5-year exposure [Larrabee 1961]. Kearny, N.J. industrial site, 15.5-year exposure. [Larrabee 19611 Curves are identified in Table 14.

the three environments. Thereafter, the reduction tapered off, gradually approaching a plateau after 0.25 percent copper content. The corrosion penetration of the steels containing copper plus a second alloying element decreased as the concentration of the second alloying element increased. This is evident from a com- and 265 and 266, shown in Figure 5 had comparable penetra- parison of the penetrations listed in Table 15(a) and from the tions. Copper and nickel were interchangeable in these amounts, corresponding curves in the copper plots. In the Kearny in- but not generally in other 1:2 proportions. dustrial environment, the addition of 1.3 percent chromium, The aforementioned conclusions apply to all three environ- 0.10 percent phosphorus, 1.0 percent nickel, or 0.52 percent ments. In general, the steels exposed in the industrial environ- silicon was most beneficial in the order listed (Fig. 5). The order ment had about the same corrosion resistance as those exposed changed in the other two environments, with 0.10 percent phos- in the rural environment and a much higher resistance than phorus being the most beneficial in semirural South Bend, and those exposed in the marine environment. The corrosion pen- 1.0 percent nickel being the most beneficial in marine Kure etration of the steels with multialloy combinations fell within a Beach (Figs. 4 and 6). Adding 0.52 percent silicon to copper 2 to 4 mils (50 to 100 m) scatterband for Mayari R weathering steels was marginally beneficial, except in the marine environ- steels that represent a broad range of compositions and expo- ment where it significantly decreased the penetration. sures in different environments [Bethlehem, undated]. See the The corrosion penetration was lowest for the high-concen- high-copper points on curves 9, 10, 11, and 12. tration multialloy combinations, as shown by curves 9, 11, and Figure 4 shows the range of copper concentrations in the nine 12. In these cases, copper concentrations in excess of 0.10 per- grades of the A588 steel. In five grades, the range varies from cent conferred no additional benefits. 0.2 percent to 0.5 percent copper. Grade D could have less Given equal concentrations of phosphorus, chromium, and copper, whereas Grades E, F, and J could have more copper. silicon, steels with 1.0 percent nickel provided the same cor- The low-copper and high-copper grades achieve a comparable rosion resistance as steel with only 0.50 percent copper. For corrosion resistance with the addition of correspondingly higher example, the underlined pairs of steels 5 and 6, 135 and 136, and lower contents of the other alloying elements. 37

Table 15. Effect of one and two-element compositions on corrosion penetration of steels. [Larrabee 1961] -Carbon 5t,.I

Penetration After Corrosion 350 KURE BEACH, N.C. Steel Concentration of Elen,entS, Percent 15.5 Years, in pm Identification first Second South Kearny, Kure 15.5 Years No.a Element I Element Bend, N.J. Beach, Pa. N.C. Rural Industrial Marine 300 (a) Effect of Copper (Figs. 4 to 6) copper t..Is I 1 0.01 Cu -- 312 732 1320 2 0.04 Cu -- 201 224 363 -- 198 201 305 3 0.10 Cu 250 4 0.24 Cu -- 163 155 284 5 0.45 Cu -- 152 135 263 35 0.48 Cu 0.20 Si 137 127 226 S 65 0.51 Cu 0.52 Si 152 117 185 208 95 0.48 Cu 0.64 Cr 130 99 r-o3 15 0.48 Cu 0.06 P 107 99 193 200 10 0.47 Cu 1.0 Ni 104 91 183 E 185 0.49 Cu 0.10 P 91 91 180 25 0.46 Cu 1.3 Cr 104 74 198 i 0 (5) Effect of Phosphorus (Figs. 7 to 9) - 150 1 -- -- 312 732 1320 Cr 91 0.06 P -- 175 198 358 I- -- 160 191 333 Ui 181 0.10 P Z 001 0.06 P 0.66 Cr 216 282 295 175 241 295 Ui 121 0.06 P 0.25 51 0. 0 lOP, 1.20 Cr, 0.58 Si 151 0.06 P 0.53 Si 173 241 241 tOO 111 0.06 P 1.3 Cr 173 221 229 II 94 0.06 P 0.21 Cu 130 124 231 2 96 0.06 P 1.0 Ni 107 104 206 OIOP, 1.20Cr, I.0Ni,053Si 95 0.07 P 2.48 Cu 107 99 193 50 (c) Effect of chromii,n (Figs. 10 toi) 1 -- -- 312 732 1320 11 0.61 Cr -- 419 1059 401 21 1.30 Cr -- 287 419 478 41 0.66 Cr 0.31 Si 351 485 381 .71 0.64 Cr 0.59 Si 216 312 201 0 0.1 02 03 0.4 05 06 101 0.66 Cr 0.06 P 216 282 295 191 0.70 Cr 0.08 P 188 249 239 COPPER CONTENT, % 16 0.61 Cr 1.0 Ni 127 137 185 14 0.63 Cr 0.22 Cu 145 117 229 Figure 6. Effect of copper content on corrosion penetration at Cr 0.48 Cu 130 99 208 15 0.61 Kure Beach, NC. marine site, 15.5-year exposure. [Larrabee

(d) Effect of nickel (Figs. 13 to 15) 1961] 1 -- -- 312 732 1320 PI4OSIORUS CONTENT IN GRADES OF A 588 STEEL. 6 1.0 Ni -- 132 155 11,16 0.5 Ni 0.60 Cr 273 598 293 21, 26 0.5 Ni 1.30 Cr 196 261 313 All Grades Except H 31, 36 0.5 NI 0.22 Si 191 253 385 61, 66 0.5 Ni 0.59 Si 170 235 243 91, 96 0.5 Ni 0.06 P 141 151 282 181, 186 0.5 Ni 0.10 P 125 147 258 4, 9 0.5 Ni 0.22 Cu 140 134 244 5, 10 0.5 Ni 0.46 Cu 135 117 223 SOUTH, BEND, PA 15.5 Years (e( Effect of silicon (Figs. 16 to 18) 1 -- -- 312 732 1320 31 0.22 Si -- 257 373 546 61 0.61 Si -- 231 361 333 71 0.59 Si 0.64 Cr 216 312 201 151 0.53 Si 0.06 P 173 241 241 81 0.54 Si 0.10 P 132 160 198 241 0.54 Si 0.10 P 132 160 198 64 0.47 Si 0.22 Cu 152 130 208 65 0.47 Si 0.51 Cu 130 117 185 E 66 0.57 Si 1.0 Ni 109 109 .152

a. The steel identification noobers are the same as those in Ref. [Larrabee 1961] 2 ti IX St,111 ti cd 7 Iii Phosphorus 0.

Figures 7, 8, and 9, respectively, show the effect of phosphorus content on the corrosion penetration in the rural, industrial, : and marine environments. -10 The corrosion penetration of steels containing phosphorus as 11 50 °15 the single alloying element is summarized in Table 15(b) and 14 plotted as curves 1 in the three phosphorus plots. In all three I rIONi_. 6 environments, adding up to 0.06 percent phosphorus to these steels significantly reduced corrosion penetration. Adding phosphorus in excess of 0.06 percent was only of marginal additional 0 cioz 0Q4 036 0.08 01 012 benefit. Phosphorus alone was more effective in reducing cor- PHOSPHORUS CONTENT, % -- Figure 7. Effect of phosphorus content on corrosion penetration at South Bend, Pa. rural site, 15.5-year exposure. [Larrabee 1961] 38

Stisi

KEARNY N.J. 'ç \ 15.5 Ysars 3J

ICURE BEACH. NC. 15.5 Ysors

2 0

L_Pho$phorus Stools

2 150-

100 .10 15 50 O I?o 14 50 0-45 Cu 13Cr lON i 050S

0-02 004 0-06 0-08 0-4 0-12 0-02 O4 0-06 0-08 0-1 0.12

PHOSPHORUS CONTENT, S PHOSPHORUS CONTENT, S

Figure & Effect of phosphorus content on corrosion penetration Figure 9. Effect of phosphorus content on corrosion penetration at Kearny, N.J. industrial site, 15.5-year exposure. [Larrabee at Kure Beach, N. C. marine site, 15.5-year exposure. [Larrabee 1961] 1961]

rosion penetration in the marine environment than in the other Chromium environments. The reduction in corrosion penetration of steels with about The effect of chromium content on the corrosion penetration 0.06 percent phosphorus and a second alloying element are given of steels in the three environments is illustrated in Figures 10, in Table 15(b), again in increasing order of corrosion resistance 11, and 12. in the industrial environment. Adding 0.48 percent copper, 1.0 The corrosion penetration of steels with chromium as the percent nickel, or 0.21 percent copper reduced penetration most, single alloying element is summarized in Table 15(c). A portion in that order. However, adding chromium or silicon made the of the chromium steel data is shown as curve 1 in Figure 10 0.06 percent phosphorus steels more susceptible to corrosion in for the rural environment, but not in Figures 11 and 12 because the rural and industrial environments, but was beneficial in the the corrosion penetration in the other environments exceeded marine environment. 16 mils (400 ,.Lm). The penetration in the rural environment In fact, the upper set of curves in Figure 8 are for steels that increased from 12.3 mils (312 m) at less than 0.1 percent contained neither copper nor nickel. Adding either one, or both, chromium to 16.5 mils (419 m) at 0.61 percent. Thereafter, it dropped the steels to the lower set of curves, in which case up decreased to 11.3 mils (287 Lm) at 1.3 percent chromium, a to 0.10 percent phosphorus only moderately increased the cor- value comparable to the penetration with only residual levels. rosion resistance. The four and five-alloy combinations in the The same detrimental effect of chromium alone, at intermediate lower set of curves (9-11, 14-16) exhibited the highest corrosion concentrations, was also observed in the industrial environment, resistance. though not in the marine environment. All A588 steel grades contain copper, and all but grade D The two-element and three-element combinations involving contain nickel. Therefore, phosphorus plays only a moderate chromium, phosphorus, and/or silicon also exhibited a peak role in increasing corrosion resistance. In fact, it is used mainly penetration at about 0.60 percent chromium when the steels for its strengthening effect (see Ch. Six, section under "Strength- were exposed in the rural and industrial environments. See ening Mechanisms"). The A588 specification limits the phos- curves 5 to 8, 12 and 13 in Figures 10 and 11. Again, this phorus level to 0.035 percent for Grade H and to 0.04 percent phenomenon was not observed in the marine environment. The for all other grades, as noted at the top of Figure 7. peaks disappeared and the corrosion resistance significantly in- 39

CHROMIUM CONTENT IN GRADES OF A5818 STEEL A - p -"C--

300 - SOUTH BEND. mium 5.5 St.els 250 Ysoru '0 059Si

O.O

0.23 Cu

10 11 50 I-ONi, O60Si

.I1ISVMIUM i.UNIII, CHROMIUM CONTENT, % Figure 10. Effect of chromium content on corrosion penetration Figure 11. Effect of chromium content on corrosion penetration at South Bend, Pa. rural site, 15.5-year exposure. [Larrabee 1961] at Kearny, N.J. industrial site, 15.5-year exposure. [Larrabee 19611

KURE BEACH, 0.595 15 5 Years

3)0

250

13 10 100 Figure 12 Effect of chromium content on corrosion penetration 0.45 Cu, 0I0p, lONe, 0.60 SI at Kure Beach, N. C marine site, 15.5-year exposure. [Larrabee 1961] 50

11NVMIUM LVNIt.NI, EX

uIR.lLL UUNTLMT IN ORAOS OF *588 STEEL KEARNY, N. J. Carbon 15.5 'Iors

350

stesI 5 I

S0K BEND, PAa 250 150 15.5 Ysars I

100

IS a- p 16 0.45 Cu. 0.IOP, 1.25 Cr,0.60 S 16 045 Cu, 0.I0P• I.25Cr, 0.60 Si

OL 0 02 04 0•6 0.8 1.2 0.2 0-4 0.6 0.8 I .2 NICKEL CONTENT, S NICKEL CONTENT, %

Figure 13 Effect of nickel content on corrosion penetration at Figure 14. Effect of nickel content on corrosion penetration at South Bend, Pa. rural site, 15.5-year exposure. [Larrabee 1961] Kearny, N.J. industrial site, 15.5-year exposure. [Larrabee 19611

creased when either copper or nickel was added, as shown by nickel. For this reason nickel was added only at the 1.0 percent the lower set of curves 2-4, 9-11, and 14-16. level [Larrabee 1961]. The penetration was lowest for the high-concentration mul- The corrosion penetration of steels with nickel as the single tielement combination (curve 16). In this instance, and for the alloying element is summarized in Table 15(d) and plotted as alloy combinations involving copper and/or nickel, the benefit curve 1 in the three nickel plots. The steep slopes show the derived from chromium increased with the content. See, for effectiveness of nickel in decreasing corrosion penetration. In example, the curves without peaks in Figure 11. Chromium was the industrial environment, for example, the penetration de- relatively most beneficial when the steels were exposed in the creased from 28.8 mils (732 m) for the carbon steel to 6.1 mils marine environment. (155 .tm) for the 1.0 percent nickel steel. The equivalency be- The chromium content of the A588 steels varies considerably. tween 1.0 percent nickel and 0.5 percent copper in achieving The high-copper steels are either low in chromium (Grades F the same corrosion resistance was discussed earlier. and H) or they have no chromium (Grades E and J). The low- The corrosion penetration of steels with 0.5 percent nickel copper Grade D composition has the highest chromium range. and a second alloying element is given in Table 15(d), in increasing order of corrosion resistance in the industrial environment. The 0.5 percent nickel level is of interest because it corresponds well with that found in many grades of A588 steel. Nickel The tabulated penetrations are the average of those for steels with 1.0 percent nickel and those with residual nickel, all other Figures 13, 14, and 15, respectively, show the effect of nickel element concentrations being the same. The addition of copper content on corrosion penetration in the rural, industrial, and as the second element was most beneficial. Phosphorus, silicon, marine environments. Previous unpublished work had shown and chromium followed copper, in that order. But none of the that the improvement in atmospheric corrosion resistance of a three was as effective as doubling the nickel content from 0.5 steel was nearly linear with increasing nickel content. In other percent to 1.0 percent. words, steels made with 0.5 percent nickel would have pene- The five-element alloy combinations resulted in the lowest tration losses about half way between those given in the three penetrations, as shown in curves 11-12 and 15-16. In these figures for steel with 1.0 percent nickel and those with residual combinations, increasing the nickel content to 1.0 percent had

IN URALM3 UI

B -

-S

Still SOUTH BEND PA. 15.5 Viors

E 3-

rl 0.IOP 023Cu )6 2 0646 U I2 1-0 Ni

50 El O44Cu,

0' 0 02 04 06 08 I 1.2 0.1 02 O V4 UD 0.6 NICKEL CONTENT, % SILICON CONTENT. %

Figure 15. Effect of nickel content on corrosion penetration at Figure 16. Effect of silicon content on corrosion penetration at Kure Beach, N.C. marine site, 15.5-year exposure. [Larrabee South Bend, Pa. rural site, 15.5-year exposure. [Larrabee 1961] 1961]

no significant beneficial effect, as indicated by the negligible But in no environment was 0.61 percent silicon as effective as, slope of the curves. In fact, this is the reason why some steel for example, 0.04 percent copper (steels 2 and 61 in Tables 15(a) suppliers do not publish a minimum nickel content only a max- and (e)). imum. When copper was omitted (curve 14, Fig. 14), nickel The corrosion penetration of steels with about 0.55 percent assumed copper's role and reduced the penetration from 6.2 silicon and a second element is given in Table 15(e). The greatest mils (157 m) at residual nickel to 2.6 mils (66 m) at 1.0 additional benefit came from adding nickel or copper, followed percent nickel. This indicates, again, that copper and nickel are by phosphorus and chromium. Generally, silicon was most ef- interchangeable from the standpoint of conferring corrosion re- fective in the marine environment, as indicated by the steeper sistance. downward slopes of the two-element alloy combinations with The nickel content of A588 steels also varies greatly. The increasing silicon content. In contrast, increasing the silicon high-copper low-chromium grades tend to be high in nickel. content did not significantly reduce the corrosion penetration The medium-copper medium-chromium grades have a lower in the rural and industrial environments. nickel content. The multielement alloy combinations were the most effective, particularly those that included copper and nickel, as indicated by curves 9-11 and 14-16. Silicon The silicon content of A588 steels is plotted at the top of Figure 16. Figures 16,17, and 18, respectively, show the effect of silicon content on the corrosion penetration of steels exposed in the rural, industrial, and marine environments. Summary The corrosion penetration of steels alloyed with silicon only is summarized in Table 15(e) and plotted as curves 1 in the Based on their weight loss study of 270 steels with systematic three silicon plots. Relative to the carbon steel, silicon was most variations of copper, nickel, chromium, silicon, and phosphorus beneficial in the marine environment in which the addition of after 15.5-year exposures in rural, industrial, and marine at- 0.61 percent silicon reduced the penetration by a factor of four. mospheres, Larrabee and Coburn [1961] concluded that: 42

400 401 Corbon Still Silicon Sts,I, KURE SEA 350 350 15.5 Ysors KEARNY. N.J. 5.5 Ysars

300 300

8 250 250 £3' z

200 200 E 3' z z 0 12 0. 3 Cu 13 ISO 150

I.- IaJz a. 100 tOO 10

50 50

0 0.1 02 03 04 05 04 °0 cii 0.

SILICON CONTENT. % SILICON CONTENT, S

Figure 17. Effect of silicon content on corrosion penetration at Figure 18. Effect of silicon content on corrosion penetration at Kearny, N.J. industrial site, 15.5-year exposure. [Larrabee 1961] Kure Beach, N.C. marine site, 15.5-year exposure. [Larrabee 1961]

Increasing the copper content of a steel from 0.01 percent EFFECT OF ALLOYING ELEMENTS FROM OTHER to 0.04 percent increased the corrosion resistance more than did INVESTIGATIONS the addition of a similar amount of any single element investigated. Next to the Larrabee-Coburn study discussed in the previous Further increases in copper were beneficial, but not to as section, Copson's [1960] work on atmospheric corrosion of steels great an extent as shown by the initial addition. is the other important source of data on the effect of alloying Relatively small single additions of nickel, silicon, or phos- elements. Copson exposed 76 steel compositions in the industrial phorus improved the corrosion resistance of carbon steel, but environment of Bayonne, N.J., and in the marine environments the largest improvements were obtained by the addition of spe- of Kure Beach, N.C., and Block Island, R.I. The steels were cific combinations of these alloying elements. arranged according to composition in 13 groups. Some were The beneficial effects of several alloying elements were not nearly free of alloying elements; others were alloyed upwards additive. Therefore, data from long-term exposure tests of many to 5 percent. The primary alloying elements were copper, phos- steel compositions were needed to determine the potential cor- phorus, nickel, and chromium. Some of the steels also contained rosion resistance of steel compositions with particular combi- molybdenum and higher than normal amounts of manganese nations of alloying elements. and silicon. Summarized herein are the data for the compositions identified in Table 16, in which the content of a single element It is apparent that various combinations of the five alloying was varied while the others remained unchanged. elements discussed herein can provide a desired corrosion re- Additional data of interest were reported by Horton [1965] sistance in a given environment. Within the bounds set forth in for a steel composition with an alloy content resembling that the ASTM specification for each grade of A588 steel, the content of A242 Type 1 steels. All steels were exposed for 17 years in of copper, phosphorus, chromium, nickel, and silicon chosen by industrial Pittsburgh, Pa. a supplier depends on: (1) the desired corrosion resistance; (2) The applicable Copson and Horton data are plotted in Figures the desired tensile and toughness properties; (3) the element 19 to 22 to examine the respective effects of copper, chromium, content naturally present in the iron ore; and (4) the availability nickel, and manganese on corrosion resistance. Additional in- and cost of the respective elements. formation on the effect of manganese, sulfur, molybdenum, and Chapter Six discusses the effect of the five aforementioned carbon was obtained primarily from two sources: (1) Chandler elements on mechanical properties. and Kilcullen's review of the corrosion resistance of low-alloy 43

Table 16. Chemical composition of steels corresponding to curves plotted in Figures 19 through 22.

Figure Reference Steel Identification No.a Composition, S No.

Cu P Cr Ni Si Mn

19 Copson 1960 26 11 2 12 16 3 varies .09 .04 .05 .03 0.38

20 Copson 1960 49 59 14 13 - - .41 .10 varies .30 .32 0.41 Horton 1965 ------0.60 0.10 varies 0.40 0.28 0.70

21 Copson 1960 32 27 - - 0.08 .01 .03 varies .19 .63 Horton 1965 ------0.60 0.10 0.60 varies 0.28 0.70

22 Copson 1960 55 40 64 56 66 17 0.28 0.10 0.3 .06 .07 varies Horton 1965 - - - - - - 0.60 0.10 0.6 0.4 0.28 varies

a. The steel identification Nos. are the smne as those in Ref. [Copson 1960].

steels in the generally more corrosive British environments Horton's data for A242 Type 1 steel further confirm the [Chandler 1970]; and (2) Boyd's review of corrosion of metals beneficial effect of chromium [Horton 1965]. Curve 7 in Figure in the atmosphere, including the weathering steels [Boyd 1974]. 20 for the industrial Pittsburgh, Pa., environment shows the The effect of copper, chromium, nickel, manganese, sulfur, same trend as the other curves. molybdenum, and carbon is summarized in the following.

Copper

Copson's data for two marine environments (curves 1 to 4) mIRE BEACH. N. C. and one industrial (curves 5 and 6) environment, plotted in 0 I 15.5'Ysars Figure 19, show a rapid decrease in corrosion penetration when 2 3.5 'r.ors the copper content is increased to 0.2 percent. Thereafter, the A BLOCK ISLAND, R.I. gain in corrosion resistance is minimal. These steels had, in 3 9.IYeors addition to copper, 0.09 percent phosphorus and negligible 4 5O'cr. BAYONNE, N.4 amounts of other elements (Table 16). The comparable steel 5 9.IYsors compositions' evaluated by Larrabee and Coburn had 0.10 per- 6 51 'or, cent phosphorus (Table 14(a), curve 10). The data from both studies correlated well. Despite some differences in exposure E conditions, both the shape and ordinates of curve 10 in Figure 5 for industrial Kearny, N.J., are quite similar to those of curve 5 in Figure 19 for industrial Bayonee, N.J. Similarly, curve 10 in Figure 6 for marine Kure Beach, N.C., compares well with curve 1 in Figure 19 at the same site. The corrosion penetration of the copper-phosphorus steels exposed in the marine environ- 0

ment was about a factor of two greater than in the industrial I SC environment.

Chromium 5 2 The beneficial effect on corrosion resistance of increasing the 6 chromium content is shown by Copson's data for the same three sites, plotted in Figure 20. The steels contained 0.41 percent Cu, 0.10 percent P, 0.30 percentNi, and 0.32 percent Si and varying amounts of chromium. The most closely related steels in the Larrabee-Coburn series contained 0.46 percent Cu, 0.10 0 0.1 0.2 0.3 0.4 0.5 0.6 percent P, less Ni (0 percent), and more Si (0.57 percent) (Table COPPER CONTENT % 14(c), curve 14). The findings from both studies again were quite similar as is evident by comparing curves 1 and 5 in Figure 20 Figure 19. Effect of copper content on corrosion penetration of with curve 14 in Ficures 11 and 12, respectively- low-alloy steels. [Copson 1960] 4 - KURE BEACH. N.C. KURE BEACH, NC. U BAYONNE. N.J. I 15.5 Years I 15.5 Years 5 9.1 Years 3SYuars 2 2 35 Years 6 5.1 Years 35 L BLOCK ISLAND.R.I. BLOCK ISLAND, R.I. O PITTSBURGH, PA. 3 9.1 Years 3 91 Years 7 17 Years 4 50 Years 4 5.OYes O KURE BEACH. N.0 BAYONNE. N.J. 8 15.5 Years 300 9.1 Years 6 5.1 Years O PITTSBURGH, Pa. 1 Il Years

E z 0 200

I- Ui z

0.25 05 0.75 1.0 1.25 .5 0.2 0.4 06 0.8 1.0 1.2 NICKEL CONTENT,% CHROMIUM CONTENT, %

Figure 20. Effect of chromium content on corrosion penetration Figure 21. Effect of nickel content on corrosion penetration of of low-alloy steels. [Copson 1960, Horton 19651 low-alloy steels. [Copson 1960, Horton 1865]

Nickel Manganese

Copson's data on the effect of nickel, plotted as curves 1 to Copson's curves 1 to 6, plotted in Figure 22, are for steels 6 in Figure 21, show the same trend as seen in the Larrabee- containing 0.28 percent Cu, 0.10 percent P, 0.30 percent Cr, Coburn data plotted in Figures 13 to 15. That is, adding nickel varying amounts of manganese and residual quantities of other has little effect when the corrosion penetration is already low. alloying elements (Table 16). They show a significant reduction The addition of nickel, however, is increasingly more bene- in corrosion penetration of medium copper steels exposed for ficial at higher corrosion penetrations. The data from both many years in marine environments, when the manganese con- sources correlated well for the 15.5-year exposures at Kure tent is increased to 0.60 percent. Beyond this level the gain in Beach, NC., given by curves 1 and 8 in Figure 21. The former corrosion resistance is not significant. Manganese had practically is for steels containing 0.08 percent Cu and 0.19 percent Si; the no effect when the copper-bearing steel was exposed in the latter for steels with 0.08 percent Cu and 0.22 percent Si (mean Bayonne, N.J., industrial environment. of steels No. 32/33 and 37/3 8). All compositions contained varying amounts of nickel. The noticeable difference between According to Horton, the corrosion resistance of A242 Type the nickel plots of the Copson and Larrabee-Coburn data is the 1 weathering steel is not affected by the manganese content shape of the curves. Based on previous unpublished data, Lar- (curve 7, Fig. 22). Evidently, in high-strength low-alloy steels, rabee and Coburn had assumed that the improvement in cor- any beneficial effect of manganese is masked by the more potent rosion resistance of a steel was nearly linear with increasing alloying elements. Boyd reported that the amount of sulfur in nickel content. For this reason, they tested at residual and 1 the steel influences the effect of manganese [Boyd 1974]. If the percent nickel contents only. Copson's nickel curves have a slight ratio of manganese to sulfur is greater than four, manganese nonlinear shape. sulfide forms. Its low electrical conductivity, he concluded, im- Horton's data for A242 Type 1 steel confirm Larrabee and proves corrosion resistance. If insufficient manganese is present, Coburn's finding that nickel has no significant beneficial effect iron sulfides of good electrical• conductivity predominate and when it is added to steels that already are alloyed with Cu, P, promote the formation of local corrosion cells. Although all Cr, and Si. This is evident from curve 7 in Figure 21 and curve grades of A588 have a manganese-to-sulfur ratio of 10, the other 12 in Figure 14, both being for steels of comparable composi- elements mask any beneficial effect of manganese, as Horton tions. had shown. 45

steel was increased from a residual amount to 0.1 percent, the o KURE BEACH.NC. 17-year corrosion penetration of the steel in industrial Pitts- I 15.5 Years 2 3.5 Years burgh, Pa., increased from 2.8 to 3.3 mils (71 to 84 m) [Horton 350 A BLOCK ISLAND, R. I. 1965]. 3 9.1 Years q L 4 5.0 Years \ 0 BAYONNE, N.J. Other Elements \ I - 9.1 Years 300- s 6 5.1 Years PITTSBUROHTUROH PA The other elements found in A588 weathering steels are carbon, vanadium, zirconium, columbium, and titanium. Carbon 250 does not appear to affect the corrosion resistance, except for possibly being harmful in high-chromium steels [Hudson 1955]. Titanium was reported to have about the same effect as nickel 200. [Otake 1962]. In fact, copper-phosphorus-titanium steels exhibited corrosion resistance superior to that of copper-phosphorus steels and copper-phosphorus-chromium steels [Muta 1965]. Vanadium (0.25 percent) steels without copper and columbium (50 were not effective [Hudson 1955]. Within the range of alloy contents specified in ASTM A588, none of these other elements contribute to the corrosion resistance of the weathering steels. I00 p

PREDICTION OF CORROSION PENETRATION

50 Legault and Leckie derived equations that allow one to estimate the 15.5-year corrosion penetration of a weathering steel with a given chemical composition [Legault 1974]. Because the r I I 00 0.25 05 075 1.0 1.25 15 equations are based on the Larrabee-Coburn data, they apply, in a strict sense, only to environments comparable to those in MANGANESE CONTENT, % semirural South Bend, Pa., industrial Kearny, N.J., and marine Figure 22. Effect of manganese content on corrosion penetration Kure Beach, N.C. Quadratic curves were fitted to the data with of low-alloy steels. [Copson 1960, Horton 19651 stepwise multiple-regression analysis techniques. The model initially consisted of five main effect terms (one for each element), ten binary-interaction terms, and five squared main effect terms. The statistically insignificant terms were eliminated. The regression equations listed below were obtained for the Molybdenum average corrosion penetration, C, at the three sites after 15.5 years exposure. The element concentration in percent must be Chandler [1970] concluded from an analysis of data reported substituted for the terms given in parentheses to obtain the by others that molybdenum may be a particularly effective al- corrosion penetration in mils. No attempt was made to extrap- loying element in British environments. For example, Edwards olate the results beyond the compositional levels of the Larrabee- [1966] had shown molybdenum to be as effective as copper over Coburn data set. a 9-year exposure period in British rural, industrial, and marine sites. On adding 0.40 percent molybdenum, the corrosion rate South Bend, Pa., rural atmosphere: was cut in half at each site. Adding 0.50 percent to 0.40 percent C (mils) = 8.50 - 13.39 (% Cu) - 3.03 (% Ni) - 21.27 molybdenum reduced the corrosion rate further in the industrial (% P) + 3.48 (% Cu % Ni) + 2.41 (% Cu % atmosphere, but had little effect in the rural and marine at- Si) + 9.55 (% Ni % P) - 1.11 (% Cr % 2 - 2.82 (% mospheres. In other tests in British atmospheres, 0.5 percent Si) + 15.31 (% Cu)2 - 0.65 (% Cr) molybdenum reduced the corrosion rate of mild steel from 5.6 Si)2 to 3.5 mil (140 to 90 m) per year [Hudson 1955]. Finally, Horikawa [1966] showed that molybdenum was more effective Kearny, NJ., industrial atmosphere: than either chromium or nickel in copper-phosphorus steels. C (mils) = 10.00 - 26.01 (% Cu) - 3.88 (% Ni) - 1.20 (% Only four grades of A588 (E, F, H, and K) contain molyb- Cr) - 1.49 (% Si) - 17.28 (% P) + 7.29 (% Cu denum in amounts up to 0.25 percent. The remaining grades % Ni) + 9.10 (% Ni % P) + 33.39 (% Cu)2 (A, B, C, D, and J) do not list molybdenum as an alloying element. Kure Beach, N. C., marine atmosphere: C (mils) = 15.49 - 16.30 (% Cu) - 4.34 (% Ni) - 4.79 (% Cr) - 12.41 (% Si) - 32.01 (% P) + 2.93 (% Cu Sulfur % Ni) + 2.46 (% Cu % Cr) + 4.36 (% Cu % Si) + 2.74 (% Ni % Si) + 12.82 (% Ni % All the evidence indicates that sulfur is harmful [Chandler P) + 1.75 (% Si % P) + 16.60 (% Cu)' + 1.20 1970]. For example, when the sulfur content of A242 Type 1 (% Cr)2 + 4.25 (% Si)' 46

Optimum additive levels of each element, within the com- kept at the lowest level, which is unlikely, these steels should positional ranges tested by Larrabee-Coburn, were determined still exhibit a 15.5-year corrosion penetration of not more than by taking partial derivatives of the predictive equations with 6.4 mils (163 .tm) in the rural, 7.1 mils (180 m) in the in- respect to each of the five additives. The optimum levels of each dustrial, and 9.3 mils (236 .tm) in the marine environments element are given in Table 17 for each exposure site. The letter tested. "a" indicates that the element concentration is the maximum The penetrations reported for some weathering steel bridges tested by Larrabee and Coburn. For the South Bend, Pa., and that are exhibiting severe corrosion exceed these values after 1 Kearny, N.J., exposure sites, only the copper concentration was year. The excessive penetration cannot be explained in terms of optimum at a level other than the highest represented in the a lean composition. It is caused primarily be a misapplication data set. For the Kure Beach, N.C., exposure site, silicon joined of the steel in unsuitable environments. copper in showing an optimum concentration at an intermediate level. All other elements reached their optimum level at the maximum concentration in the data set. The three Legault-Leckie equations were used to predict the corrosion penetration of the nine grades of A588 steel, were they to be exposed for 15.5 years in South Bend, Pa., Kearny, N.J., and Kure Beach, N.C. The range of alloy content in each grade does not always fall entirely within the corresponding Table 17. Optimum concentration of alloying elements. [Legault 1974] concentration range of the data set. To avoid extrapolation of the equations, the maximum and minimum alloy levels in any Alloying Optimum Alloy Concentration, S Element grade of A588 steel were, therefore, not allowed to exceed the South Bend, Pa. Kearny, N.J. Kure Beach, S.C. maximum and minimum concentrations in the data set. The results of the predictions are summarized in Table 18. As can Copper 0.17 0.18 0.25 be seen, all grades of steel exhibit comparable penetrations in a Phosphorus 0,12a 012a 0.12' given environment. Thus, a desired corrosion resistance can be 130a 130a 130a achieved, indeed, with several combinations of the five alloying Chromiun elements: copper, phosphorus, chromium, nickel, and silicon. Nickel 110a 110a The Legault-Leckie equations supply a useful tool for com- Silicon 064a 064a 0.42 paring the relative corrosion resistance of steels with different chemical compositions. They show also that, even if the con- Note: centration of each element in a grade of weathering steel were a. Maximtzn concentration in data set.

Table 18. Predicted 15.5-year range of corrosion penetration for A588 steels.

Predicted 15.5-Year Grade of Minimun and Maximun Concentration, S Corrosion Penetration, Mils P588 Cu P Cr Ni Si South Kearny Kure Steel Bend, Pa. N.J. Beach, N.0

005b A 0.25 0.01 0.40 0.30 - 5.5 4.4 7.5 0.40 0.04 0.65 0.40 0.65 2.9 2.3 2.9 005b B 0.20 0.01 0.40 0.15 5.9 5.1 9.3 0.40 0.04 0.70 0.50 0.50 2.9 2.3 2.9

C 0.20 0.01 0.30 0.25 0.15 5.6 4.8 9.0 0.50 0.04 0.50 0.50 0.30 4.1 3.7 5.3 050b 0 0.05 0.01 0.50 0.05 6.4 7.1 7.2 005b 0.30 0.04 0.90 0.60 3.2 2.5 3.0 10b 010b E 0.75 4.6 4.8 0-500,50 0-01 0010 b 110a 8.0 0.04 : 0.30 3.9 4.2 5.5 F 0.30 0.01 010b 0.40 0,10b 4.9 4.1 050a 9.1 0.04 0.30 1.10 0.30 3.7 4.0 4.9

H 0.20 0.01 0.10 0.30 0.25 5.5 4.8 8.7 060a 0.35 0.035 0.25 0.60 3.3 2.6 4.1

J 0.30 0.01 010b 0.50 0.30 4.6 3.7 050a 7.2 0.04 0.10 0.70 0.50 4.0 3.9 5.2 b K 0.30 0.01 0.40 0.05 0.25 5.3 4.1 7.7 0.50 0.04 0.70 0.40 0.50 3.6 3.1 3.9

Notes:

Concentration limited by maximum value in data set. Maximun concentration in P588 specification is higher.

Concentration limited by minimum value in data set. Minimum concentration in P588 specification is lower. 47

CHAPTER FOUR

ATMOSPHERIC CORROSION

Chapter Four examines the atmospheric corrosion penetration give exposure to as large an area of the underside of the spec- data for weathering steels ideally and boldly exposed in rural, imens as possible. Maximum exposure to the sun can be obtained industrial, and marine environments. Corrosion performance by exposing specimens facing south (for the northern hemi- under conditions that deviate from bold exposure are covered sphere) at an angle equal to the latitude of the test site. Exposure in Chapter Five under "Service Corrosion." at this angle will yield the lowest corrosion rate. For this reason, All values of corrosion penetration are per exposed side (sur- most American data on atmospheric corrosion of weathering face) of the specimen. steels are based on an exposure angle of 30 deg from the horizontal, facing south. Most continental European data are based on 45 deg, and most British data on 90 deg (vertical) exposure TEST METHOD angles. However, in special instances, it may be desirable to orient the racks in the direction of a specific corrodent source, The Standard Recommended Practice for Conducting At- such as an industrial plant or the ocean, or at a specific exposure mospheric Corrosion Tests on Metals, ASTM Designation G50, angle. Where it is important to obtain corrosion rates involving sets forth the general procedures that should be followed in a microenvironment, such as bridge, specimens should be conducting any atmospheric exposure test of metals and alloys. mounted directly on the structure. These tests are commonly used to evaluate the corrosion The ground beneath the exposure racks should be kept free resistance of weathering steels. of weeds, bushes, and debris. Organic herbicides, defoliants, or Multiyear exposure periods should be considered to minimize pesticides should not be used for this purpose. the influence of the variability and complexity of weather effects The steel test specimens should not be smaller than 4 by 6 and the industrial and natural factors influencing the atmo- in. (100 by 150 mm), nor thinner than 0.030 in. (0.75 mm). For spheric corrosivity of a test site. Recognizing that the corrosivity reliable results, triplicate or more specimens should be used for may vary at a site from season to season, the Standard rec- multiple removals at each exposure period. A suggested removal ommends that specimen exposures be made either at the same schedule is 1, 2, 4, 8, and 16 years. time of the year, to minimize variability, or that these differences ASTM Standard Practice Gi recommends procedures for should be established by multiple exposures. Preparing, Cleaning and Evaluating Corrosion Test Specimens. Control specimens of a composition having established weath- According to this standard, a suitable procedure for preparing ering characteristics should also be exposed along with weath- the specimens for testing might include: (1) degreasing in an ering steel specimens. The common control specimens are made organic solvent or hot alkaline cleaner; (2) pickling in an ap- of steels containing 0.02 percent max. copper and those con- propriate solution if oxides or tarnish are present; (3) abrasion taining 0.20 to 0.22 percent copper. They are hereafter referred of the steel surfaces with a slurry of an appropriate abrasive, to as carbon and copper (copper-bearing) steels, respectively. or with an abrasive paper; (4) removal of edge burrs; and (5) The Standard's recommendations ensure uniformity of test rinsing and drying. practice and eliminate extraneous variables. The resulting ex- Any mill scale or rust should be removed from all ferrous posure conditions are, therefore, somewhat reproducible but not specimens unless it is specifically desired to perform the test typical of bridge service conditions. Summarized below are the with the mill scale intact. Pickling with inhibited acid as well principal recommendations for test site location, exposure rack as blasting with sand or grit are acceptable descaling methods. fabrication, test specimen size, specimen preparation, and cal- At the end of an exposure interval, the corrosion products culation of corrosion penetration. Suggestions for special types are removed from the surfaces of the specimens either by elec- of exposure are also included. trolytic cleaning or by chemical cleaning. Electrolytic cleaning Test sites should be chosen at a number of locations repre- may result in re-deposition of metal, such as copper from re- sentative of the environment where the steel is likely to be used. ducible corrosion products and, thus, lower the apparent mass For general testing of a new steel composition's corrosion re- loss. sistance, the selection should include sites representative of rural, The average corrosion penetration per side is calculated from industrial, and marine environments. the mass lost during the time of exposure as follows: Exposure racks should be located in cleared, well-drained areas such that the exposed specimens will be subjected to the C=K (22) full effects of the atmosphere at the location of the test site. AD Shadows of trees, buildings, or structures should not fall on the specimens. Local contaminants of the atmosphere should be wheie: avoided, unless the specific influences of such conditions are intended to be assessed. The test racks should be designed to K = conversion constant; 48

C = average corrosion penetration per side, in m; Environmental Protection Agency (EPA) which require scrub- W = mass loss, in grams; bing of the effluent gases from such combustion. The critical A = total surface area (all sides), in cm2; and relative humidity, above which steels are attacked, drops to D = density, in g/cm3. between 50 and 60 percent when the air is contaminated with hygroscopic pollutants that deposit on the steel where they The constant is K = 104 when the other variables are substi- attract and retain moisture. In the absence of moisture and tuted in the indicated units. The density of low-alloy steels is condensing conditions, sulfur oxides are not necessarily aggres- D = 7.85 g/cm3. sive toward steel.

Marine Environments TYPES OF ENVIRONMENTS The losses from corrosion are greatest in a marine environ- This chapter examines the data reported in the literature for ment. Airborne salt spray droplets, salt fog, and nightly tem- test panels exposed to the atmosphere in macroenvironments perature drops can keep the steel damp for long periods in broadly classified by human activity as being rural, urban, in- certain environments. On drying, salt crystals hygroscopically dustrial, or marine. Unless noted otherwise, the data reported attract moisture at lower relative humidities than the 80 percent come from specimens exposed at an angle of 30 deg from the level found by Vernon. In fact the critical humidity is that horizontal, facing south. approaching the vapor pressure of a saturated solution of the salt [Bukowiecki 1966]. Thus, visible condensate in the form of droplets is not necessary for corrosion to proceed. The aggres- Rural Environments siveness of a marine environment is dependent on the nature of wave action at the surf line, prevailing wind direction, shoreline Rural environments are usually free of aggressive agents. topography, and relative humidity. Corrosivity from marine Their principal corrosives consist of moisture, oxygen, relatively salts decreases rapidly with increasing distance from the shore small amounts of sulfur oxides (SO), and carbon dioxide (CO,) [LaQue 1951, Hein 1977, ASTM 1968]. Severe storms can carry from various combustion products. Ammonia (NH 3), resulting salt spray inland as much as 10 miles [Ambler 1953]. Insofar from the decomposition of farm fertilizers and animal excrement as the weathering steels are concerned, the presence of chlorides may also be present. Steels rust significantly when the relative inhibits the role of sulfur oxides in forming a protective oxide humidity exceeds a certain value. For clean air this value is film. Information gained from the performance of the weath- about 70 percent. However, moisture-saturated air produces ering steels in marine exposure tests, and from structures built little rusting when dust and pollution are minimal. Some forms in coastal areas, aid in determining their behavior when con- of particulate, such as silica particles, cause no rusting, whereas taminated with roadway deicing salts. those of ammonium sulfate are very aggressive. Rural environments generally are not aggressive toward steel.

Corrosiveness of Atmospheric Test Sites Urban Environments During the periods 1951-1956 and 1960— 1962, ASTM Com- Urban environments contain sulfur oxides and nitrogen oxides mittee G- 1 annually exposed carbon steel and zinc specimens from the combustion of fossil fuels and from automobile exhaust at a number of atmospheric test sites around the world, although gases. Both contaminants promote the corrosion of steels. Be- primarily in the United States, to determine their relative cor- cause they are found also, to a larger degree, in industrial en- rosiveness toward these specific metals and the effect of periodic vironments, many studies combine urban and industrial variation in weather conditions on the corrosion rate [ASTM environments into one. This approach is taken herein. 1968]. In the first study there were eight sites in the United States and eight in Canada. The second study was expanded to include an additional 19 sites in the United States, five in Eng- Industrial Environments land, four in Panama, and one each in the Phillipine Islands and Australia. Several sites are of particular interest because The most potent causes of corrosion in industrial environ- they are the same as those for which weathering steel data are ments are the sulfur oxides (SOs) and nitrogen oxides (NOr) examined in this chapter. They are: Pittsburgh and Bethlehem, produced burning automotive fuels and fossil fuels inpowerstations Pa.; the Battersea district in London, England; Newark and [Evans 1972a, Boyd 1974, Fletcher 1979]. The sulfur oxides Bayonne, N.J.; Kure Beach, N.C.; and Point Reyes, Calif. dissolve in films of condensed moisture on steel surfaces where Table 19 summarizes the average corrosion penetration per they can be further oxidized to form sulfuric acid that attacks side for the carbon steel specimens after 2-year exposure periods the steel. Sulfur oxide also can be windbome great distances during both studies. Also given is the rating of each site relative before droplets condense out on a steel surface. Such transport to the corrosion penetration measured at State College, Pa., a is termed "acid rain." Atmospheric corrosion is most severe rural site where the data were most nearly reproducible over where the sulfur oxide levels are high and nightly temperature many years. It is apparent that a wide diversity in atmospheric drops are sufficient to result in condensation. Steels rust more corrosiveness exists between the various sites as well as weather- rapidly during the heating season when the production of sulfur related significant yearly variations at a given site. Two extreme oxides reach a maximum. However, the sulfur oxide emission examples among the American sites illustrate these points. First, levels are decreasing since the advent of the regulations by the at the 80-ft (24-rn) lot in Kure Beach, N.C., the severe marine 49

Table 19. Relative corrosiveness of atmospheres. [ASTM 19681

Ranking Location Type of 1-Year Exposure 2-Year Exposure Environment Penetration Rating Penetration Rating- (jim) (urn)

1 Norman tlls, N.W.T., Canada Rural .45 .02 3.0 .07 2 Phoenix, Az. Rural 6.60 .26 9.2 .20 3 Saskatoon, Sask., Canada Rural 6.17 .25 11.4 .25 4 Esquimalt, Vancouver Island, B.C., Canada Rural Marine 11.3 .69 26.7 .53 5 Detroit, Mich. Industrial 23.0 .93 28.9 .63 6 Fort Ihoidor Pier, Panama, C.Z. Marine 14.8 .69 29.2 .65 7 Mrenci, Mich. Rural 26.7 1.07 39.2 .85 8 Ottawa, Ont., Canada Urban 25.1 1.00 39.5 .86 9 Potter County, Pa. Rural 21.8 .88 41.1 .89 10 waterbury, Coon. Industrial -- -- 45.2 .95 11 State College, Pa. Rural 25.1 1.00 45.9 1.00 12 FiDntreal, Quebec, Canada Urban 32.5 1.30 47.1 1.02 13 Melbourne, Australia Industrial 34.1 1.37 52.2 1.14 14 Halifax (York Redoubt), N.S. Urban 31.7 1.27 53.4 1.16 15 Durham, N.H. Rural 35.4 1.42 54.7 1.19 16 Middletown, Ohio Semi-Industrial 36.2 1.44 57.6 1.25 17 Pittsburgh, Pa. Industrial 42.8 1.70 61.3 1.33 18 Columbus, Ohio Urban 41.1 1.65 65.8 1.43 19 South Bend, Pa. Semi-rural 39.9 1.60 66.6 1.45 20 Trail, B.C., Canada Industrial 48.5 1.95 69.5 1.51 21 Bethlehem, Pa. Industrial 55.1 2.21 75.3 1.54 22 Cleveland, Ohio Industrial 53.1 2.13 78.2 1.70 23 Miraflores, Panama, C.Z. Marine 55.9 2.29 86.0 1.87 24 London (Battersea), England Industrial 57.2 2.25 94.6 2.06 25 F'bnroeville, Pa. Semi-Industrial 57.2 2.29 97.9 2.13 26 Newark, N.J. Industrial 72.4 2.90 102 2.21 27 Manila, Philippines Marine 66.6 2.67 108 2.34 28 Limon Bay, Panama, C.Z. Marine 51.0 2.10 125 2.70 29 Banne, N.J. Industrial 127 5.08 155 3.37 30 East Chicago, Ind. Industrial 111 4.43 169 3.68 31 Cape Kennedy, 0.8 km from the ocean Marine 41.1 1.65 173 3.75 32 Brazos River, Tex. Industrial-Marine 107 4.30 187 4.06 33 Pilsey Island, England Industrial-Marine 89.3 3.60 206 4.48 34 London (Stratford), England Industrial 156 6.25 223 4.85 35 Halifax (Federal Building), N.S. Industrial 137 5.50 227 4.95 36 Cape Kennedy. 54 m from the ocean. 18 m elevation Marine 61.3 2.46 263 5.72 37 Kure Beach, N.C., 240 m lot Marine 85.1 3.41 292 6.35 38 Cape Kennedy, 54 m from the ocean. 9 m elevation Marine 70.8 2.84 330 7.17 39 Daytona Beach, Fla. Marine 209 8.36 592 12.9 40 widness, England Industrial 398 15.95 716 15.5 41 Cape Kennedy, 54 m from the ocean, ground level Marine 191 7.66 884 19.2 42 Dungeness, England Marine 302 12.2 979 21.3 43 Point Reyes, Ca. Marine 315 12.6 1004 21.8 44 Kure Beach, N.C., 24 m lot Marine 712 28.5 1070 23.3 45 Galeta Point Beach, Panama, C.Z. Marine 652 26.2 1382 30.2 46 Cape Kennedy, Beach 1057 42.5 -- --

atmosphere was 117 times more aggressive than the desert at- The observed aggressiveness of the marine sites greatly varied mosphere in Phoenix, Ariz. Secondly, the wide variations in with, the direction of prevailing winds, rainfall, fog, temperature, rainfall, temperature, and fog that characterize Point Reyes on and the yearly variation of these factors. With the exception of the Pacific Ocean in northern California made this site 11.5 the seaside site at Galeta Point Beach, Panama, tropical loca- times more corrosive during a 2.year period in 1960-1962 than tions are not as aggressive toward the steel specimens. There during a 2-year period a decade earlier. Variations at other sites are explanations to account for the severity of several of the were much lower. American marine sites. For example: Brazos River, Tex., is near The ASTM study suggested that rural atmospheres are less a chemical complex; Point Reyes, Calif., has variable rainfall corrosive than industrial atmospheres and much less than ma- and frequent salt fog; and Cape Kennedy, Fla., and the 80-ft rine atmospheres. But, this broad classification by macroenvi- (24-rn) lot in Kure Beach, N.C., are subjected to salt spray. ronment is not necessarily an accurate description of the Local industrial activity, traffic density, yearly changes in aggressiveness of the atmosphere towards steel. weather patterns, and the exact location of the exposure rack Examining the industrial sites, one finds that Detroit, Mich., can result in a misleading measure of the actual aggressiveness Middletown, Ohio., and Pittsburgh and Bethlehem, Pa., are of the atmosphere at a particular geographic location. The fol- known for heavy industrial activity. Nevertheless, the effects of lowing examples are cited in the study. At' Daytona Beach, Fla., local weather conditions and the location of the test sites made installation of new racks shielded some specimens on existing the atmospheres only mildly corrosive by comparison with the racks and reduced their corrosion rate. At Halifax, Canada, the Battersea and Stratford districts in London, and Pilsey Island effluent gases emitted by the smoke stack atop the Federal and Widness, England. The heavy concentration of industry in Building suggested the area to be unduly aggressive. Indeed, Newark and Bayonne, N.J., and East Chicago, Ill., is reflected the measured corrosion penetrations were 2.5 to 4 times higher by the relatively high corrosiveness of their atmospheres. than at another Halifax (York Redoubt) exposure site. At Es- 50

quimalt, Vancouver Island, prevailing winds kept the sea spray miles (50 km) north of Bethlehem, Pa. [Cosaboom 1979, Town- away from the specimens, suggesting that the marine area was send 1982]. Three steels were examined: (1) Bethlehem's A242 relatively rural despite being located a few hundred feet from Type 1, commercially available as Mayari R steel; (2) Bethle- the ocean. In Cape Kennedy, Fla., rip-rap rock that stabilizes hem's A588 Grade B, commercially available as Mayan 1-50 the beach caused the waves to disperse into large airborne water steel; (3) a 0.21 percent copper (copper-bearing) steel; and (4) droplets which do not travel far inland, yet severely corrode a carbon steel (0.021 percent Cu) similar to that used for pur- specimens exposed near the rocks. poses of determining the corrosiveness of various atmospheric In summary, in rural environments zinc and carbon steel test sites in an ASTM study [ASTM 1968]. As expected, the specimens performed consistently from year to year. In indus- weathering steels had a higher corrosion resistance than the trial environments their performance was affected by the vari- copper and carbon steels, with A242 performing better than ations in corrosiveness of the microclimates. In some marine A588. environments the degree of corrosion penetration depended Figure 24 shows corrosion penetration data after I to 5 years greatly on initial time of exposure, prevailing wind direction, exposure at three rural sites in England: (1) Loudwater, Bucks, storm activity, and distance from the shore. a rural area adjacent to Loudwater viaduct M40 with nearby The corrosiveness of any exposure site depends mainly on: light industry; (2) Silverdale, Lanes, a rural area about 3.1 miles (1) the time of wetness of the specimens, determined by the (5 km) inland from Morecombe Bay; and (3) Brixham (Mc- length of time during which the relative humidity exceeds a Kenzie 1978, Kilcullen 1979). The test specimens, exposed ver- critical value; (2) airborne sulfur oxide pollution; and (3) air- tically in the open facing the prevailing wind direction, were: borne chloride pollution. Several other aggressive factors exist A588 Grade A, commercially available as Cor-Ten B Steel; but they are less important than the three cited herein. A242 Type 1, commercially available as Cor-Ten A Steel; Unfortunately, the atmospheric variables were not recorded and (3) a proprietary steel containing 0.22 percent Cu, 0.27 during exposure testing of the weathering steels in most Amer- percent Ni, and 0.12 percent Cr plus other alloying elements. ican studies and some European studies. Methods of measuring The latter was labeled "copper" steel in Figure 24 and is not the time of wetness of steel surfaces have only recently been the equivalent of the copper steels used in American studies. developed [ASTM 1982]. The lack of climatic data prevents the The weathering steels corroded at a rate generally lower than quantitative correlation of the corrosiveness of various sites. the "copper" steel at the three rural sites. The degree of cor- Therefore, one can only infer the corrosion resistance of a steel rosion penetration appeared to correlate with the atmospheric at a new site from its known behavior at other sites. One can, sulfur content, but not with the other environmental factors. however, present the available data on the atmospheric corrosion Indeed, Table 20 shows that the atmosphere at Loudwater con- rates of weathering steels at various sites that are broadly class- tained more sulfur oxides than at Silverdale. It was also more ified by macroclimate as being rural, industrial, or marine. This corrosive to both "copper" and weathering steels. is done in the following sections. When using the data to estimate Burgmann and Grimme reported the results of 8-year tests the corrosion performance of weathering steel highway bridges, of nine steel compositions exposed at various sites in the Federal one should carefully consider the effects of microenvironment Republic of Germany [Burgmann 1981]. Of these steels, three associated with local topography, climate, and service condi- satisfied the chemical requirements of A588 Grade A and Grade tions, as illustrated by the data in Chapter Five. B steels for copper, chromium, nickel, and silicon, the primary alloying elements required for enhanced atmospheric corrosion Recent Studies resistance. One composition was a carbon steel control with 0.02 percent copper. The data for the Olpe rural site, plotted in Figure The more recent studies have increasingly focused on quan- 25, show about the same corrosion penetration as was observed titative and predictive methods of assessing corrosion resistance. for the weathering steels exposed in Saylorsburg, Pa., Figure This includes the classification and calibration of atmospheric 23. corrosion test sites [Knotkova 1982b, Baker 1982, Townsend Bonnarens and Bragard summarized the results of atmo- 1982], the evaluation of the effects of microclimate on corrosion spheric corrosion tests that were perfomed at 12 sites in Belgium, [Haynie 1982], the prediction of long-term atmospheric corro- France, Germany, Holland, Italy, and United Kingdom [Bon- sion from short-term experimental data [Bragard 1982], and the narens 1981]. The test specimens, exposed at an angle of 45 deg measurement of time-of-wetness of the specimen [Sereda 1982]. facing south, were made of: (1) Fe 37 carbon steel; (2) Fe 52 Several of these studies attempted to correlate the degree of HSLA steel; and (3) two weathering steels, respectively, with corrosion with the relevant climatological data. Application of 0.38 percent and 0.54 percent copper, 0.108 percent and 0.119 the predictive models to the large data base from previous ex- percent phosphorus, 0.48 percent and 0.49 percent nickel, 0.69 posure tests, with at least partial reconstruction of the climate percent and 0.0 percent chromium, and 0.28 percent and 0.03 from weather records, may in the future greatly enhance the percent silicon. Figure 26 shows the corrosion penetration band ability to quantitatively predict the corrosion resistance of for these steels exposed 4 years at the following three rural sites: weathering steels. (1) Ban, Italy; (2) Eupen, Belgium; and (3) Saint Germain, France. All steels exhibited lower corrosion rates in Ban than in Eupen and Saint Germain. In general, the corrosion pene- RURAL ENVIRONMENTS tration data correlated with the level of sulfur oxides in the atmosphere and with the latitude. Ban's latitude is about equal Figure 23 shows data on the average penetration per side to the exposure angle, a condition that maximizes exposure to following 1 to 8 years exposure in Saylorsburg, Pa., at Bethlehem the sun. The other two sites have higher latitudes and generally Steel's rural.test site in the Pocono Mountains located about 31 shorter drying times than those at the southern European sites. 51 I COPPER' STEEL, LOUD WATER 2 A588 GRADE A, LOUD WATER 3 COPPER" STEEL, SILVERDALE SAYLORSBURG, PA. 4 A588 GRADE A, SILVERDALE I CARBON STEEL 5 COPPER" STEEL, BRIXHAM 2 COPPER STEEL 6 A588 GRADE A,BRIXHAM E 3 A588 GRADE B 3-. 7 A242 TYPE I . BRIXHAM 4 A242 TYPE

I 111T1T: E 0 3 3' z 0 2 I- 3 ztuJ Iui a- 0 5 tO EXPOSURE TIME, YEAR

Figure 23. Corrosion penetration of carbon, copper-bearing, and weathering steels exposed to the Saylorsburg, Pa., rural environment. [Townsend 1982, Cosaboom 1979] 5 0 5 EXPOSURE TIME, YEAR Figure 24. Corrosion penetration of copper-bearing and weathering steels exposed to British environments. [McKenzie 1978, Kilcullen 19791

OLPE. F.R.G. I CARBON STEEL 2 WEATHERING STEEL E 3'

0 Figure 25. Corrosion penetration of carbon and weathering steels exposed to the Olpe, Federal Republic of Germany, rural environment. [Burgmann 1981] 0. loll /

EXPOSURE TIME, YEAR

Average Average Yearly % of Year that Average Relative Atmospheric Atmospheric Rainfall Relative Hunidity Temperature Rating Table 20. Environmental con- Site Sulfur Chlorides Was ditions during first year of test- Compounds >70% >80% of A588 ing and relative rating (mg S03/dm2/day) (mg Cl 7dm2/day) (nun) (5) (5) ('C) Grade A steels at British sites. [McKenzie 1978] Silverdale (rural) 0.30 0.03 1050 85.2 68.0 8.7 1.00 Portishead (industrial) 0.41 0.04 834 85.0 62.4 9.6 1.20 Eastney (marine) 0.49 0.15 481 76.4 53.2 10.5 1.20 Loudwater (rural) 0.48 0.03 575 82.4 64.8 8.8 1.34 Middlesbrough (industrial) 1.10 0.05 653 85.1 61.5 9.0 1.76 Tinsley (industrial) 2.05 0.07 687 74.6 47.7 10.1 2.68

Fe52 Weathering Steels

E M.

Figure 26. Corrosion penetration / bands for Fe 37 carbon, Fe 52 HSLA, and weathering steels exposed to rural environments in Bari, Italy; Eupen, 0 50 5 0 5 Belgium; and Saint Germain, France. EXPOSURE TIME, YEAR [Bonnarens 1981]

Figure 27 summarizes the corrosion penetration data for weathering steels exposed to several rural environments in the RURAL United States (curves 5 and 7), Federal Republic of Germany (curve 6), and England, (curves I to 4). These curves were replotted from Figures 23 to 25. Type of steel and exposure site are identified in Table 21. MAYARI R The corrosion penetration data for the three British sites had 18 -YEAR SCATTER a factor of two scatter and fell higher than the Saylorsburg, Pa., BAND data. The gradual reduction in corrosion rate indicates that the same corrosion mechanism that forms a protective rust film is at work in both countries, albeit at different rates. The higher corrosiveness of the British sites may be due to the longer time of wetness. Indeed, McKenzie reported that the length of time during which the relative humidity exceeded 70 percent (critical humidity for clean air) was 82.4 percent of the exposure time at Loudwater and 85.2 percent in Silverdale. In other words, during more than 80 percent of the time, the air was sufficiently humid to ensure that a condensed film of moisture on the steel surface would persist [Evans 1972a]. Given that the relative I humidity was comparable at both British sites, the atmosphere EXPOSURE TIME, YEAR with the higher sulfur oxide content was the more corrosive. Figure 27. Corrosion penetration of weathering steels exposed in The severity of attack did not correlate with yearly rainfall or rural environments. [Townsend 1982, Cosaboom 1979, Kilcullen temperature. 1979, McKenzie 1978, Burgmann 19811 The seven weathering steel curves in Figure 27 were compared with the scatterband of 2 mils to 4 mils (50 .im to 100 .tm) corrosion penetration of Mayari-R steels with a broad range of chemical compositions that were exposed for 18 years in different environments (Bethlehem undated). Extrapolation of the Table 21. Description of corrosion penetration curves for weathering rural curves to 18 years suggests that the Saylorsburg and Olpe steel exposed in rural environments (see Fig. 27). curves for weathering steel would remain inside the 18-year scatterband, but those for the British sites would rise above the Curve No. Type of Steel Exposure Site band. The corrosion penetration band for Bari, Eupen, and Saint 1 A588 Grade A Loudwater, U.K. Germain (Fig. 26) was not plotted in Figure 27. But, extrapo- 2 A588 Grade A Silverdale, U.K. lation of the data to 18 years, with the best-fit equations reported 3 A588 Grade A Brixham, U.K. in Bonnarens [1981], indicates that the corrosion penetration at Bari would fall inside Bethlehem's 18-year scatterband. Those 4 A242 Type 1 Brixham, U.K. for Eupen and Saint Germain would exceed the 18-year band. 5 A588 Grade B Saylorsburg, Pa. The corrosion resistance of the weathering steels, in compar- 6 Weathering Steel Olpe, F.R.G. ison with the the copper and carbon steels, is discussed under 7 A242 Type 1 Saylorsbarg, Pa. "Corrosion Resistance," later in this chapter.

300 I' 1 I / BAYONNE, N.J. I I. CARBON STEEL COPPER STEEL LOW-ALLOY STEEL 200 I I O2 E z I - 0 I - I- I- zw LLia- --- huh -

I -

I I I I I • I I I I I I I I 0 15 20 EXPOSURE TIME, YEAR Figure 28. A verge corrosion penetration of several carbon, copper, and low-alloy steels exposed to the Bayonne, N.J., industrial environment. [Copson 19601

INDUSTRIAL ENVIRONMENTS Copson reported the results of atmospheric exposure tests of 76 steels exposed in the industrial environments of Bayonne, N.J., and in the marine environments of Kure Beach, N.C., and Block Island, R.I. [Copson 1960]. The steels were arranged according to composition in 13 groups. Some were nearly free of alloying elements, in others the alloy content ranged upwards to 5 percent. The principal alloying elements were copper, phosphorus, nickel, and chromium; though some of the steels also E contained molybdenum and higher than the usual amounts of 3' manganese and silicon. Presented herein are the results for the steels that had chemical compositions typical of carbon steel (No. 1, 10, 44, and 45), 0.22 percent to 0.24 percent copper steels (No. 41 and 43), and the steels that contained copper, phosphorus, nickel, and chromium levels comparable to those in the weathering steels (No. 3, 13, 42, 48, 53, 66, 119 and 121). Figure 28 shows the average penetrations for the three aforementioned classes of steels that were exposed for 18 years(1941 - 1959) at the industrial site, a laboratory yard in Bayonne, N.J. Corrosion was rapid during the initial stage of exposure. There- after, the corrosion rate of the low-alloy and copper steels decreased with time, but that of the carbon steels continued unabated. Figure 29 shows Horton's data for two A242 Type 1 steels, or • , i I i I a 0.22 percent copper steel and a carbon steel (0.02 percent Cu) 0 5 10 [Horton 1965]. They were exposed in Rankin, a severe industrial EXPOSURE TIME, YEAR atmosphere in the Pittsburgh, Pa., area. The weathering steels Figure 29. Corrosion penetration of carbon, copper, and weath- again developed a protective rust film, whereas the carbon and ering steels exposed to the Pittsburgh, Pa., industrial environment. copper steels continued to corrode at appreciable rates. [Horton 1965] 54

______CARBON STEEL 200 COPPER STEEL A588 GRADE B A242 TYPE BETHLEHEM,PA. N.J. E z 0 4 I00 _..--O-2 _A..L._. w w N p - - 0. r ,- _23 I/ p 4 p 4

I C. 0IL0 L5 10 0 IJ EXPOSURE TIME, YEAR

Figure 30. Corrosion penetration of carbon, copper, and weathering steels exposed to the Bethlehem, Pa., and Newark, N.J., industrial environments. [Cosaboom 1979, Townsend 19821

A third set of data was obtained from exposures at the fol- ering steel, relative to that of carbon steel, is discussed under lowing industrial sites: (1) Bethlehem, Pa., 1.9 miles (3 km) "Corrosion Resistance." from Bethlehem's integrated steel making facility; and (2) New- McKenzie exposed the same copper and A588 Grade A steels, ark, N.J., atop a two-story building occupied by the New Jersey cited in conjunction with Figure 24 for rural environments, also Department of Transportation, near heavily traveled express- to the following three urban/industrial environments in Eng- ways [Cosaboom 1979, Townsend, 1982]. The three steels were land: (1) Tinsley, Yorks, a highly industrial area of Sheffield the same as those exposed in Saylorsburg, Pa. (Fig. 23). The adjacent to Tinsley viaduct Ml; (2) Middlesbrough, Cleveland, data from these two industrial sites are shown in Figure 30. an industrial area near the Bilhiñgham ICI works; and (3) Por- The weathering steels showed only minor variations in corrosion tishead, Somerset, an urban area with some light industry, about penetration at the two sites. By contrast, the copper steel and 1.9 miles (3 km) from the coast [McKenzie 1978]. The specimens particularly the carbon steel showed marked variations. The were mounted vertically and boldly exposed to the prevailing carbon steel specimen exposed in Newark, N.J., exhibited a wind direction. The corrosion penetration data, plotted in Figure corrosion rate much lower than in the ASTM tests (Table 19). 3 1(a), show the weathering steel corroding at a lower rate than The low corrosion rates of the carbon and copper steels suggest the copper steel at the three sites. However, the 5-year corrosion a possible error of mistaken steel identification. The carbon steel penetration of the weathering steel varied greatly from 100 behavior was also atypical of those exposed in Bayonne, N.J., at Portishead to 220 pm at Tinsley. McKenzie's statistical eval- and Pittsburgh, Pa. (Figs. 28 and 29), although their chemical uation of the relationship between corrosion rate and environ- compositions were quite similar. This can be explained, in part, mental conditions suggested that the level of atmospheric sulfur by the fact that Bayonne was a very severe environment. compounds was the main influence causing the variations in Likewise, the Rankin environment near Pittsburgh was a locally penetration. Indeed, the atmospheric sulfur level (Table 20) severe environment by comparison with that of Newark. correlated well with the industrial data plotted in Figure 3 1(a), The three sets of industrial exposure data raise questions about as well as with the rural data plotted in Figure 24. Atmospheric the adequacy of the ASTM specification requirements which chlorides, at the levels of 0.03-0.07mg Cl- /dm2 /day,measured state that weathering steels have enhanced atmospheric corro- at the sites, did not have a significant effect. Although there sion resistance of approximately two times that of carbon struc- were signs of reduction in corrosion rate with time, McKenzie tural steels with copper and four times that of carbon structural concluded that at none of the sites could the corrosion rate be steel without copper (0.02 percent max. Cu). If the corrosion considered negligible, as the British Steel Corporation had sug- resistance is based on the ratio of average penetration at the gested in a brochure entitled "Weathering Steel in Bridge end of the exposure time, the weathering steels satisfied the Work." requirements in Pittsburgh, but not in Bethlehem and Newark. An additional set of British data comes from a joint study In Newark, where the corrosion penetration was lowest, the by the British Steel Corporation and the British government's A588 Grade B steel had ironically less than two times the Transport and Road Research Laboratory [Kilcullen 1979]. In penetration of carbon steel. The corrosion resistance of weath- this study, specimens of Cor-Ten A Steel (A242 Type 1) and 55

Cor-Ten B steel (A588 Grade A) were exposed in: (1) Battersea, I.COPPER STEEL, TINSLEY 7COPPER" STEEL, TEESSIDE an industrial site; and (2) Teesside, a severe industrial site. The 2. A588 GRADE A, TINSLEY 8.COPPERSTEEL, BATTERSEA environmental conditions, angle of exposure, and chemical com- 3'COPPER"STEEL, MIDDLESBROUGH 9A588 GRADE A, TEESSIDE position of the two weathering steels and of the control steel 4. A588 GRADE A MIDDLESBROUGH tO A588 GRADE A, BATTERSEA were not reported. Presumably, the latter steel was the same as 5.COPPER" STEEL, PORTISHEAD I I. A242 TYPE I • TEESSIDE in Figure 3 1(a). The corrosion penetration data, plotted in Fig- 6 A588 GRADE A. PORTISHEAD IaA242 TYPE I • BATTERSEA ure 31(b), were similar to those for the Middlesbrough and Portishead sites. But, for unexplained reasons, the control steel 7/ corroded much faster at Teesside than at any of the four other / sites. Kilcullen and McKenzie concluded that the protective / properties of the rust film took quite a long time to develop, in I some cases many years, and the extent of development of the I rust depended very much on the atmospheric conditions of I exposure. Consequently, they further concluded, it is difficult / to predict the long-term performance of these steels even on the I basis of tests carried out over a number of years. I' E /J3 9 There is some support for this conclusion when one considers 10 a high-strength low-alloy steel composition, USS Mariner Steel, z 0 11 used in the form of sheet piling in sea water. This product I- contains 0.5 percent Cu, 0.5 percent Ni, and 0.1 percent P and 4 I /I I- exhibits enhanced atmospheric corrosion resistance in the splash I,J z 12 and atmospheric zones. Experience indicates that it requires LUa- about 7 years before the corrosion rate deviates sharply from 6 that of carbon steel. Burgmann exposed the same weathering and carbon steels, cited in conjunction with Figure 25 for the Olpe rural environment, also (1) atop a 3-story building in urban Duisburg-Huck- ingen; and (2) atop a building on the grounds of a steel rolling plant in industrial Muellheim/Ruhr [Burgmann 1981]. The (b) times of exposure were 16 and 8 years, respectively. The data, plotted in Figure 32, show that the corrosion penetration of the I I carbon steel was comparable to that of the carbon steel exposed EXPOSURE TIME, YEAR in Pittsburgh, Pa. (Fig. 29), whereas the penetration of the weathering steels was 2 to 3 times higher than in Pittsburgh. Figure 31. Corrosion penetration of copper and weathering steels At the German sites, the corrosion penetration of the weathering exposed to industrial atmospheres in U.K [McKenzie 1978, Ku- steels after 8 years was 1.6 to 2.3 times greater than that of the cullen 19781 carbon steel. The investigators concluded that: (1) the corrosion penetration during the first 8 years declines exponentially with time; (2) in severe industrial environments, the penetration increases as a linear function of time after 4 years, meaning that smallest 3-year penetration, that of the A588 Grade A steel in it does not level off; (3) changes in atmospheric pollution can Merklingen. The following can be concluded: (1) the A588 significantly change the microenvironment and the corrosion Grade A steel performed slightly better than the HRL 37 steel; rate; (4) the classification of the severity of environments as (2) the order of rating of both steels was about the same at the rural, urban/industrial, and marine is not satisfactory; and (5) six sites; and (3) the ratings did not correlate with the sulfur careful attention must be given to industrial environments in content in the rust, nor with the type of climatological data coastal cities where the simultaneous effect of sulfates and chlo- recorded at nearby 1.2, 3.1, and 12.5 miles (2, 5, and 20 km) rides may accelerate corrosion. It is a fact, however, that roof- weather stations. The atmospheric sulfur content, which was top exposures introduce variables not present at ground levels, found to correlate with penetration at the English sites, was not such as exposure to building vents, etc. recorded at the German sites. Surprisingly, both weathering The band No. 4, shown in Figure 32, envelops the corrosion steels rated only average at the Trockau site, where the 53.2 penetration data from an ongoing German study [Hein 1981]. percent mean relative humidity was much lower than at any Control specimens of two weathering steels, Resista HRL and other site. The corrosion penetration did not correlate with A588 Grade A, were exposed at six sites at some distance from yearly rainfall, sunshine, or temperature. It must be recognized freeways. The specimens were mounted at 45 deg and 90 deg from all of the other exposure data that 3 years is not a suffi- from the horizontal, facing south. Hem's data were showing ciently long period for coating A588 steel to develop a protective comparable scatter and somewhat smaller penetration after 3 oxide coating. years exposure than Burgmann's data. As all steels had similar There are two interesting sets of exposure data for urban/ alloy contents, the difference in corrosion penetration from site industrial environments in California [Reed 1982]. Instead of to site is more likely caused by variations in prevailing atmo- being made of standard thin-gage sheet, the specimens consisted spheric conditions rather than chemical composition. Table 22 of butt-welded plates, intermittently welded angles, all-welded rates the two weathering steels at the six sites relative to the angles, and mechanically attached angles. They were fabricated 56

/ / / I

/ / I CARBON STEEL, MUELLHEIM/RUHR 2 WEATHERING.STEEL, DUISBURG- HUCKft 3 WEATHERING STEEL, MUELLHEIM/RUHR E / 3.. / 3 4 RESISTA AND A588 GRADE A SCATTER / BAND FOR 6 SITES 0 4 100 //______Id z a-ILl

EXPOSURE TIME,YEAR Figure 32. Corrosion penetration of carbon and weathering steels exposed to industrial atmospheres in the Federal Republic of Germany. [Burgmann 1981, Hein 19811

Table 22. Environmental conditions and relative rating of A588 Grade A and HRL 37 steels during third year of testing at German sites. [Hein 1981]

Site Distance Sulfur Yearly Mean Yearly Mean Relative Rating from Content Rainfall Relative Sunshine Temperature Weather in Rust Hwiidity Station of A588 A588 HRL 37 Grade A Grade A (km) (%) (inn) (8) (hrs.) (C)

Merklingen 20 1.41 1158 81.4 1535 6.6 1.00 1.18 Karlsruhe 5 2.20 696 75.8 1714 10.2 1.16 1.25 Rendsburg 5 1.46 856 83.5 1471 7.2 1.23 1.38 Trockau 2 0.96 751 53.2 1451 6.2 1.24 1.38 Cologne 5 1.56 884 73.2 1352 9.7 1.32 1.54 Duisburg 5 2.03 889 77.4 1289 10.0 1.36 1.43

from one A7 carbon steel and from the following weathering deviation) than in Sacramento (1.19 mils ± 0.14 mils (30 steels: A242 Type 1, and A588 Grades A, G, and H (Mayan ± 3.6 .tm)). The Commerce and Sacramento data were R, Cor-Ten A and B, Hi Strength A, and Kaisalloy SO CR). plotted in Figure 34 and labeled No, 18 and 19, respectively. The specimens were exposed for 13 years: (1) on the grounds The low corrosion pehetration at both sites is a result of the of the Caltrans Laboratory, California Department of Trans- dry climate which prevents corrosion from occurring despite portation, on the southwestern outskirts of Sacramento; and (2) the presence of sulfur and chloride contaminants. in Commerce, an industrial area 6 miles southeast of the Los The same steels, cited in conjunction with Figure 26, were Angeles Civic Center. Both sites have low rainfall. also exposed at the following five European industrial sites: (1) The average sulfur content in the corrosion products of the Milan, Italy; (2) Liege, Belgium; (3) Stratford, U.K.; (4) Dues- weathering steels exposed in Commerce was slightly higher than seldorf, Germany; and (5) Delft, Holland [Bonnarens 1981]. in Sacramento (0.90 percent versus 0.80 percent). The chloride The corrosion penetration bands for the three steels, plotted in content was tenfold higher (0.55 percent versus 0.05 percent). Figure 33, show that corrosion penetrations of all steels exposed The average corrosion penetration was, accordingly, higher in in the industrial environments were consistently higher than Commerce 2.05 mils + 0.11 mils (52 .tm ± 2.8 tm) standard those of the same steels exposed in the rural environments, 57

Fe 37 Fe 52 Weathering Steels

NAZ

E /A' z / 2 100 I A

a. A' Figure 33. Corrosion penetration bands for I / ly Fe 37, carbon Fe 52 HSLA, and weathering I / !i steels exposed to industrial environments in Delfi, Holland; Düsseldorf FR. G.; Liege, OV Belgium; Milan, Italy; and Stratford, UK. 0 5 0 [Bonnarens 19811 EXPOSURE TIME, YEAR

Figure 34. Corrosion penetration of weathering steels exposed to industrial environments. [Copson 1960, Horton 1965, Cosa boom 1979, Townsend, 1982, Kilcullen 1979, McKenzie 1978, Reed 1982]. See Table 23 EXPOSURE TIME, YEAR for identWcation of curves.

Figure 26. In comparison with the other four industrial sites, parable chemical compositions. Curves 1 to 10 are for exposure the Fe 37 and Fe 52 steels exposed in Milan corroded most, sites in England and Germany. Curves 11 to 19 are for exposures whereas the weathering steels corroded least. The atmosphere sites in the United States. The data band for the other five in Milan had by far the highest sulfur oxide content. It appears European sites (Fig. 33) were not plotted so as not to overcrowd that the combination of high sulfur oxide and low moisture Figure 34. The band falls between curves 9 and 12. As the steels content in the Milan atmosphere was less corrosive to weath- used were about the same, the wide differences in corrosion ering steel than the combinations of lower sulfur oxide and penetration must be attributed to environmental variables of higher moisture content at the other sites. which relative humidity, diurnal temperature ranges, and at- All weathering steel data from Figures 28 to 32 are compared mospheric contamination with sulfur oxides are the most im- in Figure 34. The type of steel and exposure site corresponding portant. It is not possible to conclusively correlate the corrosion to each curve are summarized for convenience in Table 23. Also penetration with the environmental conditions because weather shown is the 18-year scatterband for Mayari-R weathering steels. and pollution data for most exposure sites are lacking. Only The steels are either of the A242 and A588 type, or have com- McKenzie, Hem, and Reed have reported such data. Hein also 58

Table 23. Description of corrosion penetration curves for weathering sites (Fig. 33) shows that the Milan curve alone would fall within steel exposed to industrial environments (see Fig. 34). Bethlehem's 18-year scatterband; the other four would exceed it after 6 to 13 years exposure. Curve No. Type of Steel Exposure Site The A242 steels exhibited a higher corrosion resistance than A588 steels exposed at the same site. 1 A588 Grade A Tinsley, U.K. 2 A588 Grade A Teesside, U.K. 3 A588 Grade A Battersea, U.K. 4 9242 Type 1 Teesside, U.K. MARINE ENVIRONMENTS 5 9588 Grade A Middlesbrough, U.K. Duisburg-Hacklngen, F.R.G 6 Weathering Steel Copson exposed at two marine sites the same steel compo- F.R.G 7 Weathering Steel Muellhein/Ruhr, sitions given in conjunction with Figure 28 [Copson 1960]. One 8 9242 Type 1 Battersea, U.K. site was in Block Island, R.I., on a bluff 15 miles (24 km) from 9 A588 Grade A Portishead, U.K. 10 Weathering Steel the mainland overlooking the open ocean; the other in Kure and A588 Grade A Six sites in F.R.G. Beach, N.C., on low ground about 800 ft (240 m) from the 11 Low-alloy Bayonne, N.J. ocean. The corrosion penetration data, plotted in Figure 35, 12 9242 Type 1 Pittsburgh, Pa. indicated that the more highly alloyed steels had better corrosion 13 9588 Grade B Bethlehem, Pa. resistance than the copper and carbon steels. The rate of cor- 14 9242 Type 1 Pittsburgh, Pa. rosion decreased somewhat over the first four years of exposure. 15 9588 Grade B Newark, N.J. Thereafter, it remained constant or increased. There were no 16 9242 Type 1 Bethlehem, Pa. signs that the rates for the copper and more highly alloyed steels 17 9242 Type 1 Newark, N.J. were tapering off, even after 17 years of exposure. The Block 18 9242 Type 1 and A588 Island site was more corrosive than the Kure Beach site. Both Grades A, G, and H Sacramento, Ca. marine sites were more corrosive to the copper and weathering 19 9242 Type I and 9588 Grades A, G, and H Connerce, Ca. steels than the Bayonne, N.J., industrial site, but they were less corrosive toward the carbon steel (Figs. 28 and 35). Evidently, the performance of carbon steel is not necessarily indicative of the performance of low-alloy steels at the same site. Additional corrosion penetration data for specimens exposed at the 800-ft (240-rn) lot, Kure Beach, N.C., were gathered by reported the concentrations of ammonia, nitrate, and chloride Cosaboom, Townsend, and Schmitt [Cosaboom 1979, Townsend in the runoff water collected from his specimens. However, other 1982, Schmitt 1969]. The former tested the same steels that had than yearly rainfall and mean temperature, two secondary var- been exposed to the Bethlehem, Pa., and Newark, N.J., indus- iables, the types of data collected do not match. The following trial environments (Fig. 30). As shown in Figure 36, all four conclusions can be drawn: steels exhibited linear or accelerated corrosion rates after the initial years of exposure, a behavior similar to that reported by The corrosion penetration was much higher at the Euro- Copson for the Block Island, R.I., and Kure Beach, N.C., sites. pean sites than at the American sites. Schmitt identified his steels as being structural carbon, copper, The primary reason for the higher penetrations at the and A242 Type 1 weathering steel, but the chemical composi- English sites appeared to be the combination of sulfur pollution tions could not be established. Each of Schmitt's steels exhibited and the longer period during which the relative humidity and, low corrosion rates, which were decreasing with time, as shown therefore, the time of wetness exceeded 70 percent. by the data plotted in Figure 37. The degree of corrosion increased with the atmospheric Comparing the Kure Beach, N.C., data reported by the three sulfur;oxide level. aforementioned investigators, one can see that the corrosion Despite the relatively high chloride and sulfur oxide con- penetration of the copper and weathering steels can vary by a centrations found in the rust in Commerce, Calif., the dry cli- factor of about two. By contrast, the carbon steels show much mate greatly limited the corrosion penetration. larger variations than the weathering steels, and appeared to be The differences in corrosion behavior at the German and extremely sensitive to the environment and to yearly variations American sites are difficult to explain precisely, because at- in climate at the same site. mospheric pollution data lack for the German sites, and cli- Figure 38 shows corrosion penetration data for two British matological and pollution data lack for the American sites. sites: one in Eastney, Hants, on a narrow promontory in Lang- Since the specimens exposed at 10 of 19 sites were fabri- stone Harbour; the other in Rye [McKenzie 1978, Kilcullen cated from A588 Grade A steel, one can assume that compo- 1979]. The steels were the same as those that were exposed in sition was not a significant variable affecting the large observed the industrial environments shown in Figure 31. It seems sur- variations in corrosion resistance. prising, at first, that the corrosion penetration of the weathering The extrapolated curves 12 to 17 would fall within Beth- steels at the two marine sites were about equal to the lowest lehem's 18-year scatterband. The curves for the European sites, penetrations for the same steels exposed at the industrial sites. 1 to 10, exceed the 18-year upperbound curve after 1.5 to 6 Compare for example the Portishead curve 6 in Figure 31 with years. Curve 11 for the Bayonne, N.J., site exceeds it after 10 the Eastney curve 2 in Figure 38. But the environmental con- years. ditions, listed in Table 20, suggest that both sites may indeed Extrapolation of the corrosion curves for the five European be equally corrosive. Both sites had comparable atmospheric

Ill ,9 3 / / / / / P / r / / 200 / E XCI z. / 0 I / I-cc ///'

/ BLOCK ISLAND, R. I. KURE BEACH, N.0 I I CARBON STEEL 4 CARBON STEEL 2 COPPER STEEL 5 COPPER STEEL 3 LOW -ALLOY STEEL 6 LOW ALLOY STEEL

Ol I I I I 1 I i I I 0 5 10 15 EXPOSURE TIME, YEAR Figure 35. Corrosion penetration of carbon, copper-bearing, and weathering steels exposed to the Block Island, R.L, and Kure Beach, N. C, marine environments. [Copson 19601

'I I KURE BEACH N.C. / KURE BEACH, N. C. I CARBON STEEL I I CARBON STEEL 2-COPPER STEEL 2 COPPER STEEL 3 4242 TYPE 3 4588 GRADE B 4 A242 TYPE I

OO II Oc E z 0

I.- z / w 1 Q2 0. bc // 1/ A'

.1 O It, 5 0 510 EXPOSURE TIME, YEAR EXPOSURE TIME, YEAR Figure 36. Corrosion penetration of carbon, copper, and weath- Figure 37. Corrosion penetration of carbon steel, copper-bearing ering steels exposed to the Kure Beach 800-ft (240-m) lot, N. C, steel, and weathering steel exposed to the Kure Beach, N. C, marine environment. [Cosaboom 1979, Townsend 1982] marine environment. [Schmitt 1969] 60

300 EASTNEY, U.K. RYE, U.K CUXHAVEN, F.R.G. I COPPER STEEL 3 "COPPER" STEEL I CARBON STEEL 2 A588 GRADE A 4 A242 TYPE 2 & 2 WEATHERING STEEL A588 GRADE A

'I /1 / I / / / / / /

OIL 0

EXPOSURE TIME, YEAR 0 5 10 EXPOSURE TIME, YEAR Figure 38. Corrosion penetration of copper-bearing and weathering steels exposed to the Eastney, UK, and Rye, U.K., marine Figure 39. Corrosion penetration of carbon steel and weathering environments. [McKenzie 1978, Kikullen 1979] steel exposed to the Cuxhaven, F.R. G., marine environment. [Burgmann 1981]

sulfur levels. Eastney had three times the atmospheric chloride The susceptibility of steels to corrosion in marine environ- level, but 10 percent lower relative humidity. ments decreases with increasing distance from the shoreline, as The same carbon and weathering steels that were exposed to the salt spray and chloride contents of the air drop. This finding the Muellheim/Ruhr industrial environment (curves 1 and 3 in is valid for steels in general. For example, Figure 41 shows the Fig. 32) were also exposed at the Cuxhaven, F.R.G., marine corrosion penetration of carbon and copper steels that were site, 500 ft (150 m) from the seashore [Burgmann 1981]. The mounted on racks located 80 ft (24 m) and 800 ft (240 m) from data were plotted in Figure 39. The aforementioned comments the shoreline in Kure Beach, N.C. [Herzog 1960, ASTM 1968]. concerning the British data also apply to the German data; i.e., At 240 m from the shore, the corrosion penetration had dropped for the same steels, the corrosion penetration in the marine by an order of magnitude. However, after 4 years of exposure, environment was less than that found in the industrial environ- there was no indication that the corrosion rate of the copper ment. This is evident by comparing curves 1 and 2 in Figure steel was leveling off. Songa [1975] also reported a rapid decrease 39 with curves 1 and 3 in Figure 32. Relative to the industrial in corrosion penetration of specimens exposed 250 m from the environment, the corrosion penetration of the carbon steel in Belgian sea coast as compared to those exposed at the shoreline. the marine environment decreased much more than that of the ASTM's calibration study of the atmospheric corrosiveness weathering steel. of test sites showed the effect of distance from the ocean and The same steels cited in conjunction with Figures 26 and 33 elevation aboveground on the corrosion of carbon steel in Cape for exposures in rural and urban environments were also exposed Kennedy, Fla. [ASTM 1968]. As summarized in Table 24, the at the following four European marine sites: (1) Biarritz, France; test sites were: (1) 2,700 ft (800 m) from the ocean at ground (2) Rye, U.K.; (3) Cuxhaven, Germany; and (4) Den Helder, level; (2) 180 ft (54 m) from the ocean at 60 ft and 30 ft (18 Holland [Bonnarens 1981]. The corrosion penetration bands for m and 9 m) aboveground, and at ground level; and (3) on the the three steels are plotted in Figure 40. The corrosion pene- riprap rock used to stabilize the beach. The corrosion penetra- trations of all steels, after 4 years of exposure in the marine tion was modest after 1 year exposure and about 4.5 times higher environments, were about the same as those of the same steels after 2 years exposure. The most severe corrosion occurred on exposed in the industrial environments (Fig. 33), but the cor- the beach, where constant exposure to spray at low and high rosion rates were not leveling off with time. No correlation was tides consumed the carbon steel specimens in less than 2 years. found between the corrosion penetrations and the amount of At 180 ft (54 m) from the ocean, the penetration had decreased atmospheric chlorides and sulfur oxides. several-fold and was about three times larger at ground level KURE BEACH, N•C I CARBON STEEL, 24m Fe 37 Fe 52 Weathering Steels 2 COPPER STEEL, 24m 3 CARBON STEEL, 240m E 4 COPPER STEEL, 240m là

5) 0 50 5 EXPOSURE TIME YEAR Figure 40. Corrosion penetration bands for Fe 37 carbon, Fe 52 HSLA, and weathering steels exposed to marine envfronments in Biarritz, France; Cuxhaven, FR. G.; Den Helder, Holland; and Rye, UK [Bonnarens 19811 A I I I I Ii Table 24. Corrosion penetration per side of carbon steel exposed in Cape Kennedy, Fla., at various distances and elevations from the ocean. [ASTM GI 1968]

Distance from 1-Year Exposure 2-Year Exposure ocean Elevation Penetration Ratinga Penetration -RatTliga 0 5 (rn) (rn) (1pi) (jo) EXPOSURE TIME, YEAR

800 ground 41 1.0 173 1.0 Figure 41. Corrosion penetration of carbon and copper-bearing 54 18 61 1.5 263 1.5 steels exposed to the marine atmosphere at Kure Beach, N.C. 54 9 71 1.7 330 1.9 [Herzog 1960, ASTM G-1 19681 54 ground 191 4.7 884 5.1 beach ground 1056 25.8

Note: a. Based on corrosion penetration at 800-rn Site. 62

than at 30-ft and 60-ft (9-rn and 18-rn) elevation. Presumably, the winds carried the large airborne droplet, caused by waves COPPER STEEL WEATHERING STEEL breaking on the riprap rock only to low elevations. At 0.5 mile ST 52 -3 WT ST 37-3 (800 m) from the ocean, the corrosion level was comparable to that of a severe industrial site. Hein investigated the effect of atmospheric chlorides on the 3 corrosion resistance and attempted to verify the requirement of 2 the 1970 German specification for weathering steels that had 5 set a minimum 3,300-ft (1-km) distance from the shoreline for 4 building bare steel structures (Hein 1977). He exposed specimens of copper steel (0.23 percent Cu) and three low-alloy steels containing Cr, Cu, Si, and Ni at 170 ft, 330 ft, 1.7 miles, and 1u] 5.0 miles (50 rn, 100 m, 2.75 km and 8.0 km) from the shore of the North Sea, Federal Republic of Germany. The first three sites were located on Sylt Island, off the North Sea coast, the I IIi5Om fourth site in Niebuell. In addition, control specimens were O2 0 lOOm exposed in the city of Karlsruhe, atop a building subjected to 200 -13 0 2.75 km the exhaust gases emitted from the chimney of a heating plant. 4 08.0km The data, plotted in Figures 42 and 43, show a sharp drop L_RUHE in corrosion penetration between the 170-ft and 330-ft (50-rn. and 100-rn) marine sites, followed by a gradual drop to the 8 WEATHERING STEEL WEATHERING STE km urban-industrial site. The three marine sites [< 2.75 km WT ST 52 WT ST 52 - 3 (1.7 miles)] were more aggressive than the Karisruhe industrial site. The corrosion rates do not appear to be leveling off after 5 years exposure. The decreasing corrosion rates at the three marine sites correlated with the decreasing amount of chlorides found in the collected rain water that had run off the specimen surfaces (Fig. 44). The chloride contents for the first year of exposure were 170 ft, 330 ft, 1.7 miles (64.6, 24.3, and 12.9 4V ; mg/l at the 50-rn, 100-rn, and 8.0-km) sites, respectively. Heavy storms pounded the coast during the third year of exposure. This tripled , the chloride content. As a result, the 33.1 mg/l third-year chloride content of the runoff water at the 5.0 mile 0.0 (8.0-km) site was higher than the first-year content at the 330- EXPOSURE TIME, YEAR ft (100-rn) site and about one-half of the first-year content at the 170-ft (50-rn) site. Evidently, strong winds carried salt spray Figure 42. Corrosion penetration of copper-bearing and weath- far inland. Water from the specimens at the Karisruhe site had ering steels exposed to marine environments at different distances only a 6.5-mg/l chloride content. but up to 15 times greater from the shoreline. [Hein 19771

L 120

I WT ST 52- 3 WEATHERING 2 WT ST 37-3 WEATHERING 40 3 ST 52- 3 COPPER 4 WT ST 52 WEATHERING

0•5 1.0 1.5 2.0 2.5 3.0 8.0 DISTANCE FROM THE SHORELINE, km Figure 43. Five-year corrosion penetration of steels as a function of the distance from the shoreline. [Hein 19771 63

TO 1364D 50m SO

0.5 1.0 1.5 2.0 3.0 8.0 DISTANCE FROM THE SHORELINE km Figure 44. Chloride content of runoff water from specimens as a function of distance from the shoreline. [Hein 19771 sulfate content. The high-sulfate, low-chloride industrial envi- of specimen configurations and steel compositions he had ex- ronment was about as corrosive as the moderate marine envi- posed in Sacramento and Commerce [Reed 1982]. The marine ronments at 1.7 and 5.0 miles (2.75 km and 8.0 km) from the test site, also used by ASTM, is located 1,300 ft (390 m) from seashore. the ocean on the Point Reyes peninsula, 36 miles (56 km) north- Hein concluded, as did many other investigators before him, west of San Francisco. It has the highest incidence of fog of that corrosion near the seashore is a function of the chloride any area in the State. The scatterband of data, after 13 years content of the atmosphere. exposure, are labeled 11 in Figure 45. The corrosion penetration Corrosion decreases with increasing distance from the sea- in Point Reyes 4.39 mils ± 0.55 mils (112 in mean ± 14 m) shore. The distance from the shore, beyond which weathering standard deviation was about twice as high as in Commerce, steels would corrode at a rate comparable to nonmarine sites, although the rust product of the specimens exposed in Com- depends both on the weather conditions and the topography of merce retained a higher sulfur content (0.90 percent versus 0.19 the area. percent) and a higher chloride content (0.55 percent versus 0.19 Finally, Reed exposed in Point Reyes, Calif., the same types percent) than the rust on the Point Reyes specimens. The reason

EXPOSURE TIME, YEAR Figure 45. Corrosion penetration of weathering steels exposed in marine environments. [Copson 1960, Hein 1977, McKenzie 1978, Townsend 1982, Cosaboom 1979, Kilcullen 1979, Burgmann 1981, Reed 1982]. See Table 25 for identification of curves. 64

for this difference was attributed to the longer time of wetness Table 25. Description of corrosion penetration curves for weathering in foggy Point Reyes than in dry Commerce. steel exposed to marine environments (see Fig. 45). All weathering steel data from Figures 35 to 39 and Figure 42 are compared in Figure 45. The data band from Figure 40 Curve No. Type of Steel Exposure Site falls between curves 4 and 6. The type of steel and the exposure site corresponding to each curve are summarized in Table 25. I WT St 37 & 52 50-rn site, Sylt Island, F.R.G. 2 Also shown is the 18-year scatterband for Mayari-R weathering WT St 37 & 52 100-rn site, Sylt Island, F.R.G. 3 Weathering Cuxhaven, F.R.G. steels exposed in nonmarine environments. The following gen- 4 4588 Grade A Eastney, U.K. eral conclusions can be drawn: 5 4242 Type 1& #588 Grade A Rye, U.K. 6 As was already noted for industrial environments, the data Low-alloy Block Island, R.I. 7 Low-alloy Kure Beach, NC. for marine environments are stratified, with the corrosion rates B 4588 Grade B Kure Beach, S.C. being higher at the European sites (curves 1 to 5) than at the 4242 Type 1 Kure Beach, B.C. American sites (curves 6 to 11). 10 #242 Type 1 Kure Beach, B.C. 11 The corrosion rates at several marine sites did not begin #242 Type 1 & 8588 Grades A, G & H Point Reyes, Ca. to level off as they do in the industrial environments. They remained at about a constant rate established after 3 to 4 years exposure. In these environments, the protective oxide film was forming at a relatively slow rate. The variability in corrosion penetration cannot be attributed be appreciated that this country is situated between latitudes 35 to the chemical compositions of the steels because they were deg on the south to 49 deg along the Canadian border. Germany, similar for all specimens. The variations were caused by differ- in the center of Europe, lies between 48 deg and 54 deg latitude, ences in airborne chloride content and time of wetness of the and England even further north between 52 deg and 58 deg specimens. Rainfall also plays a significant role in marine latitude. Because of the lower angle of the sun, the time-of- locations. wetness at the German and English exposure sites was longer At the British and German test sites, the corrosion rates and the intensity of the drying cycle was less than at the Amer- for the weathering steels appeared to be comparable for both ican sites. Thus, the northern European coastal areas are gen- the marine and industrial environments. erally more aggressive. In conclusion, it should be noted that All but curve 10 exceed the upperbound of Bethlehem's the Japanese use weathering steel despite having extensive ma- 18-year scatterband after 2 to 11 years exposure. rine exposures with high relative humidity and much rain. They try to overcome the problem by careful design and selective use On the basis of the reported data, it appears that the persistent of a porous protective coating applied to vulnerable areas (Chap- presence of salt mist and the lack of rainfall, together with a ters 9 and 11). prevailing high relative humidity for lengthy periods, combine Careful consideration should be given, however, to those lo- to slow the formation of the protective oxide film on bare weath- cations where winter conditions are of such severity that large ering steel structures. The marine locations in the United States amounts of deicing salts are used. In such instances the appli- most susceptible to these conditions can be found mainly in the cation of protective coatings to vulnerable areas, based on past foggy areas in northern California and in portions of Oregon. experience, should be considered. Similar foggy and humid conditions exist in certain areas along the Texas gulf coast. They are less severe from New Orleans to the west coast of Florida, and from Cape Canaveral to beyond CORROSION RESISTANCE Boston. In these areas, rainfall is evenly distributed throughout the year in adequate amounts to wash boldly exposed steel The ASTM specifications A242, AS 88, and A709 state that surfaces. Furthermore, fog and high relative humidity occur the atmospheric corrosion resistance of weathering steels is ap- during a few weeks to a couple of months, not year round. proximately equal to two times that of carbon structural steel Although the geographic locations which might pose a serious with copper, which is equivalent to four times that of carbon corrosion problem are relatively few in number, no generalized structural steel without copper (Cu 0.02 percent max.). This rule should or can be made as to recommending or limiting the requirement is vague in three respects. First, it does not specify application of weathering steel in coastal areas. These must be how the corrosion resistance of the weathering and reference determined on a case by case basis. Inspection of other steel steels should be measured. Secondly, it ties the performance of structures in the vicinity of a proposed bridge site can assist in the weathering steel to that of the carbon steel, whose perform- evaluating how weathering steel may perform. These include ance varies with the nature of the environment. Thirdly, it does existing bridges and their maintenance history, utility structures, not specify the chemical composition of the carbon steel other thin-gage steel structures such as barns, mail boxes, guardrails, than to limit the maximum copper content to 0.02 percent. fencing, etc. The performance of galvanized steel structures that There are three possible ways of quantifying the relative cor- have lost some zinc at cut edges also provides clues. These are rosion resistance of various compositions of weathering steels. some of the factors engineers from the steel suppliers use in They are illustrated in Figure 46 and consist of calculating the: their investigations. Chloride candle detectors can monitor the site for chlorides. 1. Ratio of corrosion penetration of the reference steel to that To explain some of the differences in corrosion penetration of the weathering steel, C,/Ca,, after a given period of exposure of specimens exposed in Europe and the United States, it should 65

EXPOSURE TIME, YEAR Figure 46. Basis for evaluating corrosion resistance of weathering steel relative to reference steel,

Ratio of corrosion rates, AC,! AC,, over a given increment the ratio of corrosion rates, dC/dt at time t: of exposure time At; and Ratio of periods of exposure, t,,jt,, at a given corrosion dC/dt AB = '26 penetration C. dC/dt AB

The task of evaluating the corrosion resistance by the three and the ratio of exposure times at penetration C: methods is simplified by fitting time-corrosion penetration curves to the data. The curve that best fits the time-penetration t,., - - 1/B,.) (27) data is a power function [Bohnenkamp 1973, Townsend 1982] t, -

C=A tB (23) The regression coefficients A and B for all sets of rural, industrial, and marine data are given in Tables 26 to 28. As indicated by the high values of the correlation coefficient which, in its logarithmic form, becomes a straight-line relation- (r = 1 means perfect correlation), Eq. 23 represents well the ship. relationship between corrosion penetration and exposure time. The ratios of penetrations, rates, and times are summarized logC = logA + Blogt (24) in Tables 29 to 31. The first two values were calculated, using Eq. 23, at the maximum time of exposure for both the weathering In this study, the following definitions and units are used: and the reference steels. The ratio of times was calculated at the largest measured penetration of the weathering steel. In this C = average corrosion penetration determined from weight way, extrapolation of the data was avoided (Fig. 46). The values loss; of corrosion resistance of weathering steels which exceeded four t = exposure time, in years; times those of carbon steel and two times those of copper steel A = regression coefficient numerically equal to the penetration were underlined. The results suggest the following conclusions: after 1-year exposure; and B = regression coefficient numerically equal to the slope of Rural environments (Table 29)—One A242 steel exposed Eq. 24 in a log-log plot. in Saylorsburg met the ASTM criteria in five or six comparisons. The other weathering steels failed all criteria. Turning to Figure For any pair of reference and weathering steels, one can 27, one observes the paradoxical situation where, despite the calculate from Eq. 23 the ratio of corrosion penetrations at time good corrosion resistance of several weathering steels, only the one corresponding to curve 7 would be acceptable according to the ASTM criteria. Industrial environments (Table 27)—Ten of 17 weathering = '- B,) (25) steels, for which enough data were available to calculate cor-

Table 26. Regression analysis of corrosion penetration data for steels Table 27. Regression analysis of corrosion penetration data for steels exposed to rural environments, exposed to industrial environments.

Curve Type of Regression Coefficients Correlation Max. Time Curve Type of Regression Coeffclents Correlation Max. Time No. Steel Coefficient of No. Steel Coefficient of Exposure Exposure A B r (years) A B r (years)

Saylorsburg, Pa. (Fig. 23) Bayvnne, N.J. (Fig. 28)

1 Carbon 31.9 .697 1.00 8 1 Carbon 139.0 0.869 1.00 3.0 2 Copper 28.8 .602 0.99 8 2 Copper 51.2 0.439 1.00 18.1 3 4588 Grade B 27.1 .481 0.99 8 3 Low Alloy 39.8 0.394 1.00 18.1 4 4242 Type 1 23.3 .316 0.99 8 Pittsburgh. Pa. (Fig.29) Loudwater, U.K. (Fig 24a) 1 Carhon 68.9 0.665 1.00 10 1 Copper 51.7 0.600 1.00 5 2 Copper 58.8 0.601 1.00 10 2 A588 Grade A 50.7 0.494 0.99 5 3 4242 Test F 40.8 0.314 1.00 10 4242 Test H 31.3 0.337 1.00 10 Silverdale, U.K. (Fig. 24a) Bethlehmn, Pa. (FIg.30a) 3 Copper 38.7 0.601 1.00 5 4 4588 Grade A 38.4 0.471 1.00 5 1 Carbon 74.8 0.339 1.00 8 2 Copper 47.0 0.403 1.00 U Briohon, U.K. (Fig. 24b) 3 4588 Grade B 47.2 0.258 0.96 U 4 4242 Type 1 34.1 0.184 1.00 8 5 Copper 39.1 0.530 0.98 5 8 4588 Grade A 28.6 0.574 0.98 5 Newark,N.J. (Fig 30b) 7 4242 Type 1 29.1 0.505 0.96 5 1 Carbon 50.4 0.346 1.00 8 Olpe. F.R.G. (Fig. 25) 2 Copper 35.0 0.310 1.00 U 3 8588 Grade B 36.1 0.273 0.97 8 1 Carbon 36.1 0.602 1.00 8 4 4242 Type 1 28.7 0.273 0.99 U 2 Weathering 21.8 0.468 0.98 U Tinsley, U.K. (Fig. 31a)

1 Copper 65.4 0.872 1.00 5 2 4588 Grade A 71.2 0.709 1.00 5 Middlesbrough, U.K. (Fig. 31a)

3 Copper 80.0 .534 1.00 5 4 4588 Grade A 59.1 .585 0.99 5 Portishead, U.K. (Fig. 31a)

5 Copper 49.4 .562 1.00 5 6 8588 Grade A 42.0 .527 1.00 5 Teesside, U.K. (Fig. 32J,)

Table 28. Regression analysis of corrosion penetration data for steels 7 Copper 182.0 0.469 1.00 5 9 4588 Grade A 65.3 0.646 1.00 5 exposed to marine environments. 11 4242 Type 1 56.6 0.666 1.00 5 Battersea, U.K. (Fig. 31b) Curve Type of Regression Coefficients Correlation Max. Time 8 Copper No. Steel Coefficient of 76.9 0.669 1.00 5 10 4588 Grade A 57.2 0.693 1.00 5 Exposure 12 A B r (years) 8242 Type 1 49.2 0.550 1.00 5 Block Island. R.I. (Fig. 35) Mullheim/Ruhr. F.R.G. (Fig. 32) 1 1 Carbon 149.8 0.755 Carbon 68.4 0.748 1.00 - 1.00 3.3 3 Weathering 2 Copper 44.3 0.848 1.00 9.1 27.2 0.848 0.97 8 3 Low-Alloy 33.1 0.706 0.96 17.1 Coninerce, Cal. (Fig. 34) Kure Beach-, N.C. (Fig. 35) -- Carbon ------ 13 19 4588 Grades 4 Carbon 71.9 0.522 0.99 3.5 5 Copper 41.7 0.711 A,G, and H -- 1.00 15.5 4242 Type 1 -- 6 Low-Alloy 35.3 0.612 1.00 15.5 -- -- 13 Kure Ueach, S.C. (Fig. 36) Sacramento, Cal. (Fig. 34) -- Carbon -- 1 Carbon 31.7 1.459 -- -- 13 1.00 8 18 4588 Grades 2 Copper 29.2 0.770 1.00 8 3 4588 Grade B 28.0 0.621 1.00 A,G and fl -- U 4242 Type 1 -- 4 4242-Type 1 22.1 0.545 0.99 8 -- -- 13

Kure Beach, S.C. (Fig. 57)

1 Carbon 43.5 0.656 0.99 7.5 2 Copper 38.1 0.614 0.99 7.5 3 4242 Type 1 19.7 0.569 0.99 7.5 Eastney, U.K. (Fig. 38) 1 Copper 53.6 0.669 1.00 5 rosion resistance, passed the ASTM criteria at least once. Only 2 A588 Grade A 42.9 0.511 1.00 5 six steels passed it in terms of the ratio of penetrations. But aye, U.K. (Fig. 38) paradoxically, 7 of 8 weathering steels with the best corrosion 3 Copper 70.3 0.623 1.00 5 4 4242 Type 1& behavior, curves 12 to 19 in Figure 34, had less than four times 4588 Grade B 49.3 0.585 0.99 5 the corrosion penetration of their carbon steel counterparts. The Cuxhaven, F.R.G. (Fig 39) weathering steels fared better in terms of the ratio of times, with 1 Carbon 56.2 0.547 1.00 8 2 Weathering 40.4 0.512 0.88 U 10 of 15 passing the ASTM criteria. But, those passing included weathering steels with good (curves 12 to 17) and poor corrosion Point Reyes. Cal. (Fig. °)...... resistance (curves 2 and 4). -- Carbon 13 11 4588 Grades A, G, and H 3. Marine environments (Table 28)-The distribution of 4242 Type 1 13 weathering steels passing and failing the ASTM criteria is quite 67

Table 29. Corrosion resistance of weathering steels in rural environments (see Fig. 27).

Curve Type of Exposure Corrosion Resistance Based on Ratio of No.. Steel Site Penetrations Rates Times C Cu Cu C Cu

1 A588 Grade A Loudwater, U.K. ... 1.2 ... 1.5 ... 1.4 2 A588 Grade A Silverdale, U.K. ... 1.2 ... 1.6 ... 1.4 3 A588 Grade A Brixham, U.K. ... 1.3 ... 1.2 ... 1.6 4 A242 Type 1 Brixham, U.K. ... 1.4 ... 1.5 ... 1.9 5 A588 Grade B Saylorsburg, Pa. 1.8 1.4 2.7 1.7 2.4 1.7 6 Weathering Steel Olpe, F.R.G. 2.2 ... 2,8 ... 3.7 7 A242 Type 1 Saylorsburg, Pa. 3.0 2.2 6.7 4.3 4.9 3.8

Table 30. Corrosion resistance of weathering steels in industrial environments (see Fig. 34).

Type of Exposure Corrosion Resistance Based on Ratio of Curve Times No. Steel Site Penetrations Rates C. Cu C Cu C Cu

1 A588 Grade A Tinsley, U.K. ... 1.2 ... 1.5 ... 1.2 2 A588 Grade A Teesside, U.K. ... 2.1 ... 1.5 ... 4.8 3 A588 Grade A Battersea, U.K. ...... 1.2 ... 4 A242 Type 1 Teesside, U.K. ... 2.3 ... 1.6 ... 6.1 5 A588 Grade A Middlesbrough, U.K. ...... 1.1 .. 6 Weathering Steel Duisburg, F.R.G. ...... ... 7 Weathering Steel Muellheim/Ruhr, F.R.G. 2.0 ... 1.8 ... 2.6 8 A242 Type 1 Battersea, U.K. ... 1.9 ... 2.3 ... 2.6 9 A588 Grade A Portishead, U.K. ... 1.2 ... T7.3 ... 10 Weathering Steel & A588 Grade A Six Sites in F.R.G...... 11 Low-alloy Bayonne, N.J. 5.9 1.5 13.0 1.6 20.5 2.4 12 A242 Type 1 Pittsburgh, Pa. 3.8 2.8 8.0 5.3 7.3 5.5 13 A588 Grade B Bethlehem, Pa. 1.9 T 23 2Tr 14 A242 Type 1 Pittsburgh, Pa. 4.7 3.4 9.2 6.2 101 .T 15 A588 Grade B Newark, N.J. T T T T 4.1 1.2 16 A242 Type 1 Bethlehem, Pa. 3.0 2.2 5.6 4.8 26.3 6.9 17 A242 Type 1 Newark, N.J. 2.0 1.3 2.6 1.5 7.9 2.4 18 A242 Type 1 & A588 Grades A, G & H Sacramento, Ca. 1.1 ...... 19 A242 Type 1 & P588 Grades A, G & H Commerce, Ca. 1.5 ...... 68

Table 31. Corrosion resistance of weathering steels in marine environments (see Fig. 45).

Curve Type of Exposure Corrosion ResistanceBased on Ratio of No. Steel Site Penetrations Rates Times C Cu C Cu C Cu -

1 WT St 37 & 52 Sylt Island, F.R.G...... 2 WT ST 37 & 52 50 and 100-rn sites ...... 3 Weathering Cuxhaven, F.R.G. 1.5 ... 1.6 ... 2.1 4 A588 Grade A Eastney, U.K. ... 1.6 ... 2.1 ... 2.0 5 A242 Type l& A588 Grade A Rye, U.K. ... 1.5 ... 1.6 ... 1.9 6 Low-alloy Block Island, R.I. 4.8 1.8 5.1 2.2 8.9 2.3 7 Low-alloy Kure Beach, N.C. 1.8 1.5 1.6 T78 2.4 T 8 A588 Grade B Kure Beach, N.C. 6.5 1.4 15.2 1.8 3.6 1.6 9 A242 Type 1 Kure Beach, N.C. 9.6 2.1 25.7 3.0 4.7 2.6 10 A242 Type 1 Kure Beach, N.C. 2 20- 3.0 Z! _47.4 TT 11 A242 Type l& - - A588 Grades A, G & H Point Reyes, Ca. 1.9 ......

similar to that cited for the industrial environments, with the reaches the 0.010-in. (250-sm) penetration after only 8-years inconsistencies repeating themselves. For example, those passing exposure. on the basis of the ratio of times again included weathering The two curves for the Newark, N.J., and Teesside, UK., steels with good (curve 10 in Fig. 45) and poor corrosion re- environments, labeled 15 and 2 in Figure 34, are examples of sistance (curve 4). very good and very poor corrosion behavior of weathering steels. Bare weathering steel could be used in Newark, but definitely The clean air standards imposed by the Environmental Pro- not in Teesside. But Townsend's method of calculating corrosion tection Agency have gradually reduced air pollution. The gen- resistance, as the ratio of time to 0.010-in. (250-m) penetration, eral improvement in air quality appears to have benefited the makes the steel satisfy the ASTM requirements at both sites. carbon steels more than the weathering steels. For example, The question does not seem to be whether one should use the comparing the pre-1965 data in Figure 29 with the pre-1982 ratio of penetrations, rates, or times to calculate corrosion re- data in Figure 30, one notes a large drop in the carbon steel sistance. The observed anomalies come from tying the behavior curve and a small drop in the A242 Type 1 steel curve. The of weathering steel to that of copper and carbon steels. The improved air quality and the manner in which the ASTM stand- decision to build a bridge from bare weathering steel should be ards specify corrosion resistance have combined to create an based on the anticipated corrosion behavior of the weathering anomaly. That is, the advantages of the more corrosion resistant steel at the site under consideration. Data for copper and carbon steels tend to be understated in the less corrosive environments. steels help to assess the corrosiveness of the environment, but Furthermore, the ASTM approach does not consistently dis- the behavior of these reference steels should not be part of the criminate between good and poor corrosion resistance. Town- specification requirements for corrosion resistance of weathering send and Zoccola calculated the corrosion resistance as a ratio steels. Bridges would not be built from bare copper and carbon of times at what they considered to be a design allowable average steels anyway. corrosion penetration per side of 0.010 in. (250 zm) [Townsend From an engineer's viewpoint, design specifications should 1982]. Applying this method to the data for the mildly corrosive state the requirements for corrosion resistance in terms of the Newark, N.J., industrial environment (see Fig. 30) raises indeed penetration per side of the weathering steel. Two approaches the corrosion resistance of the weathering steel to 2.1 and 11.8 seem possible. In one, the projected corrosion penetration at the times those of the copper and carbon steels. But it also requires end of the service life could be limited to a value small enough extrapolation of the weathering steel data, beyond the maximum so that a plate thickness increase would not be needed to com- 8-year exposure time in Newark, to 1,260 years in order to pensate for section loss. Figure 2 of Bethlehem's Mayari R calculate the ratio of times to 0.010 in. (250 m) penetration Booklet implies this approach. It shows a scatterband of pen- [Townsend 1982]. The data for the highly corrosive Teesside, etration data for weathering steels representing a broad range U.K., industrial environment (see Fig. 31) gives an extreme of chemistries and exposures in different environments [Beth- example on the other side of the spectrum. Applying Townsend's lehem undated]. The upper-bound curve of the scatterband method to this case, gives the weathering steel more than five reaches 4-mils (100-m) penetration at 14 years and continues times the corrosion resistance of the copper steel, although it to rise at 0.05 mils (1.27 j.Lm) per year. Weathering steel bridges

designed on the basis of this information would be expected to have corrosion penetration at 50 years of service of C = 100 m + (50-14) years x 1.27 .&m/year = 150 Am (6 mils) per side. Townsend states that in practical applications of structural steels, where members are generally 2 in. (13 mm) or greater, a lO-mil (250-sm) penetration per side results in a loss of load- bearing cross section of 4 percent or less [Townsend 1982]. This loss is said to be insignificant and to fall within the design allowances. The upper-bound curve on Bethlehem's data and E Townsend's design allowance are plotted in Figure 47. a. Taking another approach, the allowable limit on corrosion z 0 penetration could be raised provided that member thicknesses I- are increased, above the values required for carrying load, to offset the loss due to corrosion. The West-German and East- German weathering steel specifications follow this approach. The West-German specifications give an envelope for weathering steels exposed to severe industrial environments [DASt 1979]. They require the following increase in minimum thickness above that needed for carrying load;

REQUIRED THICKNESS INCREASED PER SIDE, AM (mils) SERVICE LIFE LIGHT MEDIUM SEVERE (years) CORROSION CORROSION CORROSION 30 - 800 (32) 1000 (40) 60 800 (32) 1200 (48) 1500 (60) EXPOSURE TIME, YEAR

Figure 47. Guidelines for corrosion penetration of weathering The following examples of corrosiveness are given: severe— steels. industrial and marine atmospheres; medium—urban atmospheres; light—rural atmospheres. The envelope and the required thickness increase per side are plotted in Figure 47. The East German specifications set the following upper limits on corrosion penetration per side of their KT weathering steels [DDR 1983].

SERVICE LIFE ALLOWABLE CORROSION PENETRATION PER SIDE, (years) j.m (mils) Bethlehem Steel's upper-bound envelops the data for 75(3) 5 150 (6) American rural and industrial sites, but not for the marine sites. >5 150 (6) + 20gm/year (0.8 mils/year) Townsend's limit envelops the data for all American rural and industrial sites. It also envelops the A242 Type 1 curves for the Kure Beach, N.C., marine site. But this grade of weathering steel is no longer used for bridges, because the high phos- These limits are drawn in Figure 47 as a continuous curve. The phorus content reduces impact strength. specifications also require that the designer check the load- The West German envelope covers all domestic and foreign carrying capacity of the net section after corrosion losses. data for rural, industrial, and marine environments. To compare the observed performance of weathering steels The East German envelope covers all domestic and foreign against the aforementioned design guidelines, Figures 48, 49, data for rural and marine sites, but not the data for most Eu- and 50 for rural, industrial, and marine environments, respec- ropean industrial sites. tively, were constructed. In these log-log plots, the exponential curves fitted through the data shown in Figures 27, 34, and 45 Two additional factors should be considered in choosing become straight lines. The lines were drawn solidly for the length guidelines for corrosion design. First, corrosion penetration un- of exposure time and extrapolated with dashed lines to 50 years, der service conditions can be significantly higher than under the design life of highway bridges in the United States. Com- ideal conditions of bold exposure examined in this chapter. paring the guidelines for allowable corrosion penetration (Fig. Secondly, corrosion pits, which run two to three times greater 47) with the three plots leads to the following observations than the average penetration, reduce the fatigue life of some regarding the corrosion performance at the end of the 50-year details [Copson 1960; Hiam 1978]. These factors are examined design life: in Chapters Five and Seven, respectively. 70

RURAL 1000

500 400 -I 300 - 3 30C E - - - j 200 ------.- 5 20C 0 -- -s------I-4 - - - 100 - ---_------ ------ 7 - ------50 - - - - - 40 30

- I 2 3 4 510 20 30 40 5060 EXPOSURE TIME, YEAR EXPOSURE TIME, YEAR Figure 48. Projected corrosion penetration of weathering steels Figure 49. Projected corrosion penetration of weathering steels in rural environments (see Table 21 for identification of curves). in industrial environments (see Table 23 for identification of curves).

E 26 0

10 510 20 30 40 50 EXPOSURE TIME, YEAR Figure 50. Projected corrosion penetration of weathering steels in marine environments (see Table 25 for identification of curves).

CHAPTER FIVE

SERVICE CORROSION

Weathering steels develop a protective oxide coating that exposed in the same season of different years. These differences effectively shields the underlying steel base from the corrosive are greater for carbon steel than for weathering steel, meaning environment when the following conditions are met: that the more corrosion resistant steels are less affected by the initial climatic conditions. The increase in heating fuel use with Exposure to the atmosphere wherein intermittent cycles the onset of winter, and the corresponding emission of com- of wetting and drying occur without prolonged periods of damp- bustion gases, was one reason why higher corrosion penetrations ness. were observed in fall exposures. A second reason may be at- Washing of the exposed surface by rainwater to remove tributed to the fact that a semiprotective oxide film forms during contaminants. the spring and summer months. Absence of heavy concentrations of corrosive pollutants, Larrabee [1966] also cited data for carbon steel (0.04 percent especially salt spray from any source. Cu) exposed to the Kure Beach, N.C., marine environment, Absence of detail geometries that trap moisture, dirt or beginning in 1944 and 1949. The specimens faced north, east, debris and, hence, foster a corrosive environment. south, and west. They were sheltered to various degrees. In all

The service environments in which weathering steel bridges are exposed do not necessarily fulfill each of the aforementioned conditions. In many cases a bridge will still exhibit adequate corrosion resistance, whereas under severe conditions it may 300 not. FALL EXPOSURES This chapter examines the corrosion behavior of weathering B 1934 A 1931 steels under localized or microenvironment conditions more F' 1934 C,C 1934 typical of the service conditions encountered at bridge sites. The D, D' 1935 main effects to be examined are initial climate, sheltering, orientation, angle of exposure, time of wetness, atmospheric pol- E,E' 1938 lutants, salt, and debris. In addition, depending on service G,G' 1945 conditions and type of detail, the steel may be subjected to the following forms of corrosion: poultice, galvanic coupling, pitting, and crevice corrosion. The sections of this chapter summarize A relevant data available for each case. E E All values of corrosion penetration are per exposed side (sur- D face) of the specimen.

INITIAL CLIMATIC CONDITION AF

The corrosion rate of steel specimens is influenced by the climatic condition at the beginning of the exposure period. Schramm and Taylerson reported major differences in the corrosion rate of steel specimens that were exposed at the same site at successive 2-month intervals [Schramm 1934]. Ecc(D- Larrabee [1966] reported the corrosion penetration of specimens that were exposed at various times to the Kearny, N.J., D' E' industrial environment. The specimens were fabricated from a F' low-carbon steel with 0.05 percent Cu and a high-strength low- alloy steel containing 0.80 percent Si, 0.15 percent P, 0.43 percent Cu, and 1.1 percent Cr. The original exposure was in the 0 5 10 spring or fall of the same year, or it was in the same season of EXPOSURE TIME, YEAR different years. As can be seen from the data plotted in Figure 51, those specimens initially exposed in the spring, regardless Figure 51. Effect of climatic condition on corrosion penetration of the type of steel, corroded less than those exposed in the fall. of carbon and HSLA steels exposed to the Kearny, N.J., industrial The curves also are markedly different for specimens initially atmosphere. [Larrabee 1966] 72 cases, the 2-year corrosion penetration of the specimens exposed sections had been exposed to salt spray during transport from in 1944-1946 (Fig. 52(a)) was two to three times greater than Japan. those exposed in 1949-195 1 (Fig. 52(c)). These results suggest the possibility of more frequent bad weather, such as storms and hurricanes, during the first exposure period. SHELTER AND ORIENTATION During the periods 1948-1955 and 1960-1962, a Task Group of ASTM Committee G-1 calibrated the corrosiveness of the The effect of shelter on the corrosion resistance of steels can atmosphere at many domestic and foreign sites [ASTM 1968]. be examined in two types of test. In one type, the corrosion For this purpose, new low-carbon steel (0.09 percent C) spec- penetration of the skyward surface versus that of the ground- imens were exposed for 1 or 2 years, beginning in each year of ward surface is determined by painting one surface of the spec- the test periods. Figure 53 shows the 2-year corrosion penetra- imen and exposing the nonpainted surface either skyward or tion data for the following sites: (1) Kure Beach, N.C., 800 ft groundward. In the other type of test, nonpainted specimens (240 m) from the ocean; (2) semirural South Bend, Pa.; (3) semi- are either boldly exposed or sheltered from wind, rain, and industrial Middletown, Ohio; and (4) rural State College, Pa. sunshine. The first test gives the corrosion penetration of only The figure shows that the atmosphere at the rural, semirural one surface, whereas the second test yields the mean for both and industrial sites have not changed significantly during the surfaces. two test periods. The Kure Beach data illustrate the large dif- To determine the relative corrosion penetration that occurs ferences that can be caused by changes in climatic conditions, on each surface, Larrabee [1941] performed the first type of test even at the same test site. The committee speculated that the by exposing carbon, copper, and low-alloy steel specimens combination of salt spray and little rain, or widely dispersed painted on one side in semirural South Bend, Pa., and industrial periods of rain, considerably increased corrosion. Kearny, New Jersey. The chemical composition of the steels In summary, the climatic conditions at initial exposure can were not given. The angle of exposure was 30 deg facing south. produce large differences in corrosion penetration. These dif- The data, plotted in Figure 54, reveal a much smaller corrosion ferences are much larger for carbon steel than for weathering penetration of the skyward surfaces than of the groundward steel. For neither type of steel, however, does the difference surfaces. The slopes of the respective curves indicate that the disappear with increasing exposure time. Evidently, the initial rust on the skyward surface became increasingly more protective reactivity of the steel with its environment affects the long-term with time of exposure. After 3.5 years, the ratio of corrosion corrosion behavior. For this reason, initially contaminated penetration of the skyward and groundward surfaces was about weathering steel should be cleaned after installation. This was 38 percent to 62 percent for the three steels at both South Bend, done in the case of the Luling Bridge, La., many of whose Pa., and Kearny, New Jersey.

Roof Eoves

CD Ri1 \ 3mm Ii

l6SmnJ \\ 31 F]T J a

Ground

N E S W N E S W

Figure 52. Effect of orientation and sheltering on corrosion of carbon steel exposed to the Kure Beach, N. C., marine atmosphere. [LaQue 1951; Larrabee 19661 73

350 200 L 4 I Kure Beach, N.C. 2 South Bend, Pa SOUTH BEND PA. SOUTH BEND PA. 300 . 3 Middle Town, Ohio Skyward Groundword 4 State College, Pa. 150

250 I I II 100 1I 200 I' , I " / I 50 ,02 Siso E 4 I 100 / I Carbon Ste1 1 Steel - L\L2 200 3Low-A 50 _3 Lii !Lt—

Lii KEA N.J.'I KEARNY N.J. Skyward Groundward l50 946 1962 INITIAL EXPOSURE , l Figure 53. Efftct of initial year of exposure on 2-year corrosion penetration of carbon steel. [ASTM G-1 1968] 00

According to Larrabee, the most likely explanation for one 2 side corroding less than the other was the washing of the skyward surface by rain. As a result, the sulfur compounds, which 50 - are believed to be largely responsible for the severity of atmo- goD3 spheric corrosion of inland sites, were washed away to a greater go-o3 extent on the skyward than on the groundward surface. The latter became wet by dew, but were not washed by rain. A C I I I I 0 50 5 second reason for the difference was the more rapid drying of the skyward than the groundward surface. Larrabee also ob- EXPOSURE TIME, YEAR served more severe corrosion along the cut edges than on the Figure 54. Effect of skyward versus groundward exposure on flat faces. corrosion of steels in the rural atmosphere of South Bend, Pa., Additional information on the effect of shelter was obtained and the industrial atmosphere of Kearny, N.J. [Larrabee 19441 from specimens exposed vertically in groups of five each, as shown in Figure 52(d) [Larrabee 1966]. Specimens 1 and 2 were completely sheltered and specimen 3 was partially sheltered. Specimens 4 and 5 were boldly exposed to wind, rain, and were about 20 percent less corroded than the sheltered speci- sunshine. The specimens faced north, east, south, and west. The mens. sheltered test racks were located in semirural South Bend, Pa., and in marine Kure Beach, N.C. 800 ft (240 m) from the ocean. The greater amount of sunlight received by the two bottom The types of steel exposed were carbon, copper, and a HSLA specimens facing south and west reduced the time of wetness. steel of the Cu-P-Cr-Si type. This helped to form a more protective rust film than that found The 4-year data for the semirural test site (1944-1948) showed on the specimens facing north. But even the bottom specimens that the differences in the amount of shelter and direction of facing south in this test had 50 to 100 percent higher corrosion exposure caused only small variations in penetration for the losses than the duplicate control specimens facing south at a three steels. The effect of the amount of shelter in the four 30-deg angle from the horizontal. directions was as follows [Larrabee 1966]: The results were quite different for the marine atmosphere at Kure Beach, N.C. The 4-year data for carbon steel, plotted East—Practically no effect of shelter. in Figure 52(b), show the following [Larrabee 1966]: North—The boldly exposed specimens 4 and 5 were about 20 percent more corroded than the sheltered specimen 1, 2, and East—The sheltered specimen 2 lost three times as much 3. weight as the boldly exposed specimens 4 and 5. The partially South or West—The boldly exposed specimens 4 and sheltered specimen 3 lost twice as much weight. 74

North and South—The sheltered specimen 2 lost 2-2.4 Table 32. Effect of orientation of 3-year corrosion penetration of carbon times as much weight as the boldly exposed specimens 4 and steel and Czechoslovak weathering steel Atmofix 52A [Knotkova 5. 1982a]. West—The losses at all positions, sheltered and exposed, were within 20 percent. Orientation Mgle Corrosion Penetration,um Carbon Steel Weathering Steel These extreme variations in corrosion loss occurred because East 45 104 73 during the first few months of exposure hurricane-force winds West 45' 61 61 deposited more seawater spray on the specimens facing east South 45 86 40 (toward the ocean) than on those facing north and south. The South horizontal 83 53 specimens facing west received least seawater spray. The losses of the boldly exposed specimens 4 and 5 were lowest because driving rains washed off the salt. Figures 5 2(a) and (c) compare the 2-year data from the 1944 tests with data from a similar test begun in 1949 at the same This finding seems reasonable, as the bridge was not opened to site. Two important differences are apparent. The specimens in traffic for the initial 6.5 years of testing. Hence, no salt-con- the 1949 test lost about one-half as much weight as those in the taminated water could have leaked through the joints, and little, 1944 test. Furthermore, in the 1949 test, the two sheltered if any, debris would have accumulated near the abutments. specimens 1 and 2 lost less weight than the boldly exposed Zoccola performed a corrosion study of carbon and A242 specimens 4 and 5. Evidently, lower velocity winds during the Type 1 steels in Detroit, Michigan, that was similar to the 1949 test failed to drive seawater spray under the protective previously cited Newark, N.J., study [Zoccola 1976]. The spec- eaves. The specimens in all five positions lost less weight than imens were vertically and horizontally mounted on: (1) an in- the control specimens facing south at a 30-deg angle from the terior girder of Eight-Mile Road over U.S. 10; and (2) the roof horizontal. of the National Guard Armory roof, nearby the bridge. The Knotkova, et al., examined the effect of orientation on carbon bridge was opened to traffic in 1964, the freeway below in 1965. steel and Czechoslovak weathering steel, designation Atmofix The horizontally mounted specimens had been painted on one 52A (Table 10) [Knotkova 1982a]. The specimens were exposed side and were exposed with the nonpainted surface facing either for 3 years at a 45-deg angle from the horizontal, facing east, skyward or groundward. In this way, corrosion penetration data west, and south. The results of the weight loss measurements, were obtained for surfaces that simulated the top and bottom given in Table 32, showed that the eastward exposure was the faces of a girder flange. most aggressive. The reason for this behavior was attri1uted to The data, plotted in Figures 56(a) and (b) for the vertically the slower drying time of the specimens facing east. Southerly and horizontally mounted specimens, respectively, show the exposure was the most favorable orientation for both steels. It effect of shelter in a corrosive bridge environment. The bridge- also resulted in the greatest corrosion resistance for the weath- sheltered A242 specimens (solid curve 1) corroded up to 6.4 ering steel relative to that of the carbon steel. times more than those boldly exposed on the roof of the Armory In Chapter Four, section under "Industrial Environments," building (solid curve 2). In fact, when sheltered by the bridge, corrosion penetration data were reported for carbon and A242 the A242 specimen corroded much like carbon steel. The poor Type 1 steels boldly exposed in industrial Newark, N.J., at a performance of the A242 steel was due to the accumulated 30-deg angle from the horizontal [Cosaboom 1979]. These same deposits of road salts, debris, and corrosion products that kept steels also were used to examine the effect of shelter on corrosion. the surfaces wet, trapped chlorides and sulfates, and caused The specimens were mounted vertically and horizontally at accelerated poultice-type attack [Zoccola 1976]. The data in three locations: (1) on an interior girder of Bridge No. 9 on Figure 56 also show that the corrosion rate of the bridge- Interstate 1-78, Section 5V; (2) on an exterior girder of the same sheltered A242 steel specimens was increasing with time. bridge; and (3) on the roof of a nearby two-story building oc- The horizontal exposure data (Fig. 56(b)) suggest two con- cupied by the New Jersey Department of Transportation. The clusions. First, the corrosion penetration of A242 steel on the 1-78 bridge is part of a ramp. It was opened to traffic 6.5 years skyward surface was lower (46 to 54) under bold exposure, but after construction and exposure of the weathering steel speci- higher (65 to 44) under sheltered exposure. This behavior re- mens. During that time there was no traffic on or below the sulted from beneficial washing of the boldly exposed skyward bridge. surfaces by the rain, and from detrimental debris accumulation The corrosion penetration data are plotted in Figures 55(a) on sheltered skyward surfaces, respectively. Secondly, for bold and (b) for the vertically and horizontally mounted specimens, and sheltered exposures, the average corrosion penetration per respectively. In both cases, the bridge-sheltered specimens side was greater for the horizontal than for the vertical surfaces. (curves 1 and 2) corroded up to 65 percent more than those The last example on the effect of shelter presented herein boldly exposed on the building roof (curve 3). The vertically comes from British data on carbon and A588 Grade A steels mounted specimens, which simulated a web, corroded more on [McKenzie 1978]. The test sites were located in rural Iden, rural the interior girders than on the exterior girders. The horizontally Loudwater, industrial Tinsley, and marine Shoreham-by-Sea. mounted specimens, which simulated a flange, corroded equally At each test site, sets of specimens of each steel were exposed on interior and exterior girders. The results were comparable vertically: (1) under the deck of a bridge; (2) under a 12 X 12 for both carbon and A242 Type 1 steels. in. (300 x 300 mm) wide-flange beam, with the web placed There was no difference in corrosion penetration of specimens horizontally so that the lower halves of the flanges would shelter mounted on an interior girder at midspan or near the abutment. the specimens and with the upper half of the beam thermally 75

300 NEWARK, N.J. I interior Girder,! 78 2 Ixterior Girder, 178 3 Roof, N.J. Dot 250

Carbon Steel A242 Type I

200 (a) Vertical I (b) Horizontal

z o 150• ,A2

4 of 3 I— / -

a. "

3 50 Bridge Opened I...- ..-Bridge To Traffic i Opened To rroffic CO 5 10 0 5 10 EXPOSURE TIME, YEAR EXPOSURE TIME, YEAR Figure 55. Effect of sheltering on corrosion of steels exposed to the Newark, N.J., industrial atmosphere. [Cosaboom 1979]

300 DETROIT,MICH. I ISky I Interior Girder, 8 Mile Rood I 2 Roof Of Armory I Carbon Steel 250 j4242Type1 I Ground I IGround / 200 I E /

- (a) Vertical / (b)Horizontol z / 150 4 . / 2 Ground I— / / z CL IOC Sky / _...L2 2Ground

2 Sky 50 . —D 2

I I I I I I I 0 5 10 0 5 10 EXPOSURE TIME, YEAR EXPOSURE TIME, YEAR Figure 56 Effect of sheltering on corrosion of steels exposed to the Detroit, Mich., industrial atmosphere. [Zoccola 1976] 76 insulated; and (3) in the open facing the prevailing wind direc- exposure was the increase in penetration at Shoreham-by-Sea, tion. the marine site, again pointing to the severe corrosive effect of McKenzie concluded the following from the data plotted in chlorides under sheltered conditions. Figure 57. The corrosion rates measured beneath a bridge (curve Generally, in open and sheltered exposure at British sites, 1) can differ considerably from those measured in open exposure weathering steel did not perform significantly better than carbon (curve 3). At Loudwater and Tinsley, bridge-sheltered corrosion steel. rates for both steels were much lower than in open exposure. At the marine site, however, the opposite was found, indicating the adverse effect of chlorides under sheltered conditions. At ANGLE OF EXPOSURE Iden, corrosion rates under open and bridge-sheltered exposures were similar. The shapes of the penetration versus time curves Exposure of a specimen facing south (for the northern hem- differed for open and bridge-sheltered exposures. In particular, isphere) at an angle equal to the latitude of the test site yields at rural Loudwater and marine Shoreham-by-Sea, the corrosion the lowest corrosion rate for most materials [ASTM G-50 1976] rates did not decrease with time under sheltering so that, for according to the ASTM G50 Standard Recommended Practice A588 Grade A, the rate over the last 2 years of the test was for conducting Atmospheric Corrosion Tests on Metals. In the higher than in open exposure. United States, most published data on atmospheric corrosion of McKenzie further concluded that the beam shelter did not metals are based on an exposure angle of 30 deg from the adequately represent the bridge shelter. The corrosion penetra- horizontal. In comparison, central United States falls between tion under these exposures was similar at Loudwater, but not the latitudes of 30 deg in New Orleans, La., and 49 deg at the at any other site. The penetration measured for specimens be- North Dakota border with Canada. The British normally expose neath the beam depended on the position of the specimen. It specimens vertically, with the United Kingdom reaching from was found in some later tests that specimens placed vertically about 50 deg to 60 deg latitude. The Germans and Czechoslo- beneath the beam had l 2 times the penetration of those placed yaks mostly use 45 deg, their latitude being about 48 deg to 55 near the sides. This suggests variations in the microenvironment deg. and, therefore, in corrosion penetration with type of shelter. Taylerson reported that vertically exposed specimens lost The most noticeable difference in beam sheltered versus open weight somewhat faster than those exposed at 30 deg [Taylerson

I Bridge Shelter LOUDWATER, U.K. TINSLEY, U.K SHOREHAM -BY -SEA 2 Beam Shelter (Rural) (Industrial) (Marine) 3 Open Exposure A Carbon Steel 71 A588 Grade A / / / II

IDENI U.K. E (Rural) z 150 0 — CC , I- 100

/ 50 II

I I I I 101 0 5 7. 50 5

EXPOSURE TIME, YEAR

Figure 57. Efftct of sheltering corrosion on steels exposed at four British sites. [McKenzie 19781

1941]. Larrabee quotes LaQue as stating that the repro- mens and (2) in partially sheltered exposures, vertical specimens ducibility of results was poorer when the specimens were ex- are washed less and should corrode more than horizontal spec- posed vertically than when they were exposed at 30 deg [Lar- imens. Since the specimens were mounted on the 1-78 bridge rabee 1966]. He attributed this to a slight variation in the 6.5 years after it was opened, traffic could not have affected the position of the specimen from the vertical which results in a results. The differences cannot be explained with the available slight sheltering of the specimen and reduces the amount of rain information. washing that surface. Zoccola [1976] in a similar series of tests, exposed specimens The following three examples illustrate the effect of angle of vertically and horizontally on the roof of the National Guard exposure on corrosion of carbon and A242 Type 1 steels. Cos- Armory and on an interior girder of Eight-Mile Road over U.S. aboom [1979] measured the corrosion penetration of specimens 10 in Detroit, Michigan. The data for carbon and A242 Type mounted vertically, horizontally, and at a 30-deg angle. The test 1 steels from Figure 56 are replotted in Figure 59 to show the sites were on the roof of the New Jersey D.O.T. building and effect of angle of exposure. (The curves for skyward and ground- on Bridge No. 9 on 1-78. The data from Figures 30 and 55 are ward surfaces were averaged). On the roof of the Armory build- replotted in Figure 5 8(a) and (b), respectively, to show the effect ing, the vertical specimens corroded less than the horizontal of angle of exposure. The roof specimens corroded more in specimens, contrary to what one would expect under bold ex- vertical exposure than in horizontal and 30-deg exposures, as posure. They also corroded less when hung on an interior girder one would expect when specimens are rain washed and sun of the Eight-Mile Road bridge. But in this case the observed dried. Both steels exhibited this behavior. behavior was expected, because less debris and salt spray kicked The fully sheltered specimens on interior girders corroded up by the traffic below the bridge accumulated on the vertical more in a vertical than in a horizontal position. For the partially than on the horizontal specimens, owing to drainage. sheltered specimens on exterior girders, however, just the op- The third example comes from German data for A588 Grade posite was found. Both observations are contrary to the following A steel specimens that were exposed at six sites, both vertically expectations: (1) in fully sheltered exposure, vertical specimens and at 45 deg from the horizontal [Hein 1981]. The 45-deg trap less debris and should corrode less than horizontal speci- exposure data were replotted from Figure 32. The vertical cx-

Newark, N.J. I Vertical 2 Horizontal 2 3 30° Carbon Steel A242 Type I (a) Roof, N.J. 0.0.1. (b) Bridge Girder, 178

, lint. 2 tnt. z 2 Ext. 0 ISO Ext.

I- Lii z LAJ /5 lint. 2 mt. 2 Ext. 8~_ __z /7rr 4I Ext.

50 04, r'—Bridge Opened To Traffic

01 I I I I I .I I I I _I 0 5 10 0 5 10 EXPOSURE TIME, YEAR EXPOSURE TIME, YEAR

Figure 58. Effect of exposure angle on corrosion of steels in Newark, N.J., industrial atmosphere. [Cosaboom 19791

DETROIT, MICH. I Vertical 2 Horizontal 250 ---Carbon Steel A242 Type I 2

a) Roof Of. Armory (b) Interior Girder, Eight - Mile Rood,2/ E

2 401

Ee 5 10 0 5 10 EXPOSURE TIME, YEAR EXPOSURE TIME, YEAR

Figure 59. Effect of exposure angle on corrosion of steels in Detroit, Mich. [Zoccola 1976]

posure data were added in this section. Figure 60 shows that and 0.66 percent Si) [Larrabee 1966]. They were exposed in: (1) at five of six sites vertical specimens corroded more than those a railroad tunnel with continuously wet walls; (2) a second exposed at 45 deg, as one would expect for bold exposure. tunnel with dry walls; and (3) on the roof of an office building Horizontal exposure was also found to be more corrosive than in industrial East Pittsburgh, Pa. The HSLA steel specimens 45-deg exposure in 3-year tests of Czechoslovak weathering steel, exposed on the building roof (curve 3 in Fig. 61) was more Atmofix 52A (Knotkova 1982a]. corrosion resistant than the weathering steels in the industrial In summary, under bold exposure, vertical and horizontal environments corresponding to curves 1 to 15 in Figure 34. Its specimens corroded more than specimens exposed at an angle 4-year penetration was 4.4 times smaller than that of carbon close to the latitude of the site. This has been known for a long steel at the same site. In the two tunnels, the HSLA steel time. Vertical specimens were expected to corrode more than specimens corroded like their carbon steel counterparts. The horizontal specimens in bold exposure, but not in sheltered atmosphere in the wet tunnel was so severely corrosive that exposure. Some data confirm those expectations, other data do after one year of exposure the penetration of the HSLA steel not. The perceived contradictions could result from differences was 11 times that of the same steel exposed on the roof. in microclimate that were not fully understood. Direct precipitation of rain, snow, or condensation is not needed for a steel surface to remain continuously wet. Moisture can be deposited by capillary action of the porous oxide coating, and adsorption by corrosion products or salt deposits on the CONTINUOUSLY MOIST CONDITIONS surface. Pores in the oxide coating, cracks and crevices, small pits, and the space between the steel surface and a settled speck Alternate cyclQs of wetting and drying are one of the four of dust all foster capillary action. Forces of attraction between necessary conditions under which the weathering steels develop water molecules and a solid surface increase the possibility of a protective oxide coating. Under continuously moist conditions, adsorption condensation of water vapor on the surface of a solid especially with water of low pH, weathering steels do not exhibit [Tomashov 1966]. superior corrosion resistance [Larrabee 1966]. Such moist atmospheric corrosion, as opposed to wet atmo- For example, Figure 61 shows corrosion penetration data for spheric corrosion, takes place when a thin, invisible film of a low carbon steel (0.024 percent Cu) and an HSLA steel (0.36 electrolyte forms on the surface at a relative humidity well below percent Cu. 0.13 percent P, 0.08 percent Ni, 1.0 percent Cr, 100 percent. The critical relative humidity, above which the 79

lOG

E 50

fR7 R8 E

RIO 0 5 0 EXPOSURE TIME, YEAR Figure 60. Effect of angle of exposure on corrosion ofA588 Grade A steel at six German reference sites. [Hein 19811 atmospheric corrosion rate increases sharply, depends on the EXPOSURE TIME, YEAR type of metal, surface contaminants, atmospheric pollutants, and Figure 61. Effect of continuously moist conditions on corrosion the nature of the corrosion product. In the following, data on of steels. [Larrabee 1966] critical relative humidity for several combinations of steel surface conditions and pollutants are examined. Vernon [1931, 1933, 1935] first showed the effect of sulfur dioxide, in conjunction with relative humidity, on the corrosion Table 33. Water vapor pressure above saturated aqueous solution of of iron. He corroded carbon steel specimens in bell-jar atmo- salts at 20'C (687) [Tomasbov 1966]. spheres with controlled humidity, SO2 content and temperature. In the absence of SO2, even saturated humid air (100 percent Salt ZnC1 CaCl N1IN0 NaCl NHCI Na SO (NH ) SO Ccl ZnSO K So relative humidity) did not cause appreciable rusting. When the SO2 content was raised to 0.01 percent, the critical relative Relative Hunidity 10 35 67 78 79 81 81 86 91 99 humidity dropped to 60 percent. At 0.01 percent SO2 content and relative humidities of 70, 75, and 99 percent, iron corroded 7, 8, and 12 times as much as in 100 percent humid air free of SO2. The presence of 0.03 percent carbon dioxide, the normal atmospheric content, greatly reduced corrosion. dissolution in moisture adsorbed from the air, forming an elec- Vernon also found that silica particles (dust) neither appre- trolite. Most notably, the results indicate that sodium chloride, ciably increased corrosion nor lowered the critical relative hu- a widely used deicing salt, becomes moist at 78 percent relative midity. Ammonium sulfate, however, lowered the critical humidity and 20 C (68 F). relative humidity to 80 percent. Table 34 shows the effect of steel surface condition on the Table 33 gives relative humidity values of air in an enclosed critical relative humidity. The corrosion rate of iron, with a space located above saturated solutions of various salts prior oxide coating that formed in water or in 3 percent sodium [Tomashov 1966]. These values correspond to the critical rel- chloride solution, increases sharply when the relative humidity ative humidity at which the respective salts begin to wet by of the air rises above 65 and 55 percent, respectively. At these 80

Table 34. Critical value of relative humidity at which the corrosion Table 35 lists the chemical plants and the atmospheric constit- rate of iron sharply increases [Tomashov 1966]. uents. The data for specimens exposed 2 years are plotted in Figure 62. Those exposed only 6 months and one year were not Surface Condition Critical R.H. (6) plotted because reliable assessment of corrosion behavior is not possible after such a short time. The curves are identified in- Clean surface in pure air about 100 70 Table 35. Clean surface, corroding in air + 0.01% SO2 The data show that the atmosphere at the sulfur plant (curve Prior slight corrosion in water 65 1) was much more corrosive than at the other plants. The Prior corrosion in 3% NaCl solution 55 atmosphere at the chior-alkaJi plant (curve 2) was the next most corrosive, but it was appreciably less corrosive than at the sulfur plant. The atmospheres at the other plants (curves 3 to 7) were somewhat more corrosive than the industrial atmospheres at American industrial sites for which data were reported in Figure critical humidities, a continuous film of moisture forms by cap- 34. The A588 Grade A steel was generally superior to carbon illary condensation or by the hydration of salts, corrosion prod- steel, but slightly inferior to the more richly alloyed A242 Type ucts, and other films that might be present on the surface. These 2 steel. results indicate that weathering steels with a dense rust coating Knotkova, et al., tested carbon steel and Czechoslovak weath- (smaller capillaries) in a salt-free environment would likely cor- ering steel, Atmofix 52A, in the localized climate of chemical rode beginning at relative humidities somewhat higher than 65 plants, agricultural areas, metallurgical and textile production percent, while those with a loose rust coating (larger capillaries) plants, as well as in automotive and street car environments in a salt-laden environment would corrode at 55 percent. [Knotkova, 1982a]. The results of these 5-year exposures are Structural steel members sheltered from the sun and wind given in Table 36. They show that weathering steels are generally generally dry at a slower rate than exposed members. As a result not suited for such applications; most definitely not for atmo- of the high heat conductance of steel, its surface temperature. spheres containing chlorides. In localized climates, where the ), weathering steel is sig- can fall below the ambient and dew point temperatures. Both major pollutant is sulfur dioxide (SO2 factors tend to lengthen the time of wetness. However, because nificantly more resistant than carbon steel. of the massiveness of a structure, there could be a time delay Knotkova, et al., also investigated the corrosion behavior of low-carbon steel and Atmofix 52A weathering steels as a func- before condensation occurs. pollution. The results of the 5-year Forty-eight years have passed since Vernon found that cor- tion of the level of SO2 rosion of iron accelerates only beyond a critical relative hu- exposure tests at a number of Czechoslovak sites were grouped midity. The significance of this fact was not fully appreciated as follows: (a) rural and urban atmospheres with average annual /day or less; (b) urban and until Sereda developed a method for measuring the percentage SO2 pollution levels of 40 mg/m2 pollution level reached of time when this critical humidity-is exceeded [Sereda 1960]. industrial atmospheres where the SO2 This period, called "time of wetness," was subsequently shown 90 mg/m2/day; and (c) heavily polluted industrial atmospheres /day. The data to be the most important factor promoting the atmospheric where the SO2 level reached 120 to 150 mg/m2 for the three groups, plotted in Figure 63, show a progressive corrosion of metals. 2 content in the The time of wetness can be measured with a moisture-sensing increase in corrosion penetration with the SO air. The rust coating on the weathering steel had a dark brown element consisting of alternate electrodes of copper and gold, to violet color. Its structure in the industrial atmospheres was or zinc and gold, spaced from 4 to 8 mils (100 to 200 m) apart coarser than in the rural atmospheres. For heavily polluted [ASTM 1982]. The recorded output potential gives a measure of the duration of wetness on the sensing element. Experience atmospheres (Fig. 63(c)), the corrosion level was higher than has shown that the sensing element reacts to factors causing is generally seen on weathering steels. In this case, the rust wetness in a similar manner as the surface on which it is coating was spalling and regrowing, both regularly and inten- mounted. sively. This method does not relate the time of wetness to levels of Knotkova, et al., recommended a maximum limit of 90 ambient relative humidity. However, studies showed that the mg/rn 2/day on annual average SO2 pollution above which a annual time of wetness of panels exposed under standard con- weathering steel structure should be protected against corrosion. ditions is equivalent to the cumulative time during which the Above this limit the structure should be designed for loss of relative humidity exceeded a certain value. The time of wetness metal and load-bearing capacity during the service life. The of bridges at different locations can be greatly influenced by periodic shedding and regrowth of the rust coating becomes local atmospheric and service conditions. There is at present no significant, and the rust coating ceases to be protective as in- reliable method of estimating the time of wetness for various tended. However, in view of the EPA standards, the installation surface conditions from meteorological records. of SO2 scrubbers in fossil fuel plants and the use of low-sulfur coal, excessive level of SO2 are unlikely to occur in most Amer- ican industrial sites. INDUSTRIAL POLLUTANTS The effect of sulfur dioxide on corrosion was also investigated by Hayme, et al., who tested A242 Type 1 specimens in envi- An indication of how weathering steels might perform in ronmental chambers at controlled levels of sulfur dioxide, ni- industrial environments around chemical plants can be gained trogen dioxide, ozone, relative humidity and temperature from the results of short exposure tests of carbon (0.01 percent [Haynie 1978]. During the 1,000-hour exposure time, weathering Cu), A242 Type 2, and A588 Grade A steels [Schmitt 1967]. was simulated with programmed dew/light cycles of 40-mm

Table 35. Corrosion penetra- Corrosion Penetration, urn Exposure tion of carbon and weather- Curve Type of Carbon A242 A588 Time, ing steels exposed to vanous No. Plant Atmospheric Constituents Steel Type 2 Grade A Years chemical plant atmospheres (see Fig. 62). [Schmitt 1967] 1 Sulfur Chlorides, sulfur and sulfur Compounds 1100 518 823 2

2 Chior-alkali F.bisture, chlorides and lime 478 145 188 2

3 Chlorinated Chloride compounds 272 56 56 1 hydrocarbons

4 Chior-alkali M2isture, lime and soda ash 211 53 48 2

5 Sulfuric acid Sulfuric acid fumes 114 53 56 1

6 Petrochmmical Chlorides, hydrogen sulfide and sulfur 86 30 48 2 dioxide

7 Petrocheslcal Ammonia and amnonius acetate fumes 58 33 48 1

I 12 I Carbon Steel A242 Type 2 A588 Grade A

35(

30( -I I 2

25( li E

20( F—

I— w I j Z 15( LU 2 0 I I / I I I I

i 6

4 5( 4 J'I X 6 6 Figure 62. Corrosion penetration of car- V t I F hon and weathering steels exposed to var- 2 ious chemical plant atmospheres (see EXPOSURE TIME, YEAR Table 35). [Schmitt 19671

duration. The corrosion penetration of the specimens was es- The following empirical equation accounted for 91 percent sentially a linear function of time, contrary to the declining of the variability in the experimental data. power function of time normally observed in naturally weathered A242 steel. Evidently, a protective oxide coating did not 44 - 31150'\ ]t. C = 5 64 SO2 + e (55 (28 form. ,Th results of the study are of interest in the sense that RTI they showed concentration of SO2, relative humidity, and temperature to be the important factors controlling corrosion. in which:

45C Table 36. Corrosion behavior of carbon and Atmoflx 52A steels in man- o Atmofix 52A ufacturing and other specific 40( Corbon steel microclimates: (1) production of Glauber salt; (2) production of H2SO4; (3) NaCI transporter, (4) C12 bottling platform; (5) NaCl store- 30( house; (6) compressor room, NH3; (7) electrolysis of NaCl; (8) lique- fying of Cl2; (9) atmospheric test station Letnany; (10) open-cut mining excavator; (11) overburden dumping machine; (12) separator of combustion products; (13) road bridge; (14) stadium; (15) storehouse of wood; (16) winter stadium; (17) greenhouse; (18) underground coal mine; (19) cowhouse-skylight; (20) cowhouse, loft; (21) little bull house and loft; (22) lead-coating work- shop; (23) synthetic fertilizer store- iiL house; (24) textile dyeworks; and 1 2 3 4 5 67 8 9 10 11 12 13141516 17 1819202122232425 (25) textile washing works. Indoor1ik Outdoor Sheds Indoor Outdoor Sheds 2-year Test 5-Year Test

URBAN AND INDUSTRIAL HEAVILY INDUSTRIAL 40m9/m2/ Day 90 i Content90mq/m2/Doy mg/m2/ Day -----Low Carbon Steel 44 imiyr/ I / Atmof ix SOA / / I

/ 123jJm E 77/ 12)jm/Yy., " RURAL ,, S02 Content , Day / 4)Jm/Yr. / 00- --- / / / --- , - ' 2m/Yr

501..

rei 0 5 0 50 5 EXPOSURE TIME, YEAR

Figure 63. Effect ofsulfur dioxide content on corrosion of low carbon and weathering steels exposed to various atmospheres in Czechoslovakia. [Knot- kova 1982 (a)] 83

C = corosion penetration, ,.Lm; Some beams are corroding at a rate of 5 to 6 mils (125 to 150 = time of panel wetness, calculated from an empirical ,nn)/year/surface. equation, year; The cross-section loss of the upper and the lower flanges SO2 = concentration of SO2, g/m3; of beams not exposed to salt spray appears to be about the same, R = gas constant, 1.9872 cal/g-mol; and even though the upper flange is only exposed on one side. T = geometric mean panel temperature when wet, °K. The upper and lower portions of the web corroded at about the same rate, and slightly faster than the middle portion. Although the predicted corrosion penetrations agreed poorly The portions of a beam near the abutments had lower with actual values measured in cities, the results of the chamber corrosion rates than those directly over the roadway. study give valuable insight into the effect of SO2, relative humidity, and temperature. The results of the exposure tests in Newark, N.J., and Detroit, Mich., cited under "Shelter and Orientation," were replotted in Figures 64 and 65 in a manner that shows the effect of salt contamination on corrosion of weathering steel bridges [Cosa- DEICING SALTS boom 1979; Zoccola 1976]. The specimens exposed on building roofs (curve 3) indicate that the atmospheres at both locations Deicing salts spread on highways during the winter can create were comparable, with Newark being marginally more corrosive. a more corrosive environment for the bridge than that encoun- The corrosion penetration of the weathering steel specimens tered at marine sites (Chapter Four "Marine Environments"). mounted on the interior girders of the 1-78 bridge in Newark On heavily salted highways in the snow-belt region of the United (curves 1 in Fig. 64) was 1.6 times that of their roof counterparts. States, the salt contaminates the steel structure in the following This increase was caused by sheltering, not salt, because the ways: (1) salt water leaks through the bridge deck, mainly at bridge was not opened to traffic during the first 6.5 years. The expansion and construction joints, and drains longitudinally corrosion penetration of the weathering steel specimens mounted along sloped bottom flanges; (2) a mist of saltwater runoff is on the interior girders of the Eight-Mile Road Bridge in Detroit kicked up in the wake of trucks passing beneath the bridge and (curves 1 in Fig. 65) was, on average, 4.4 times that of the settles on the steel structure; (3) dust containing salt particles specimens on the Armory roof. The difference between the two from dry roadways is blown against the steel structure; and (4) ratios, 1.6 for 1-78 and 4.4 for Eight-Mile Road, shows the effect salt water lying on the upper surface of the bottom flange wicks of salt that existed at the latter but not at the former bridge. up the web by capillary action as much as 10 in. (250 mm). In fact, at the Eight-Mile Road Bridge, the weathering steels Salt accumulates on sheltered surfaces that are not washed specimens corroded like the carbon steel specimens. by rain. Quantitative data are available from the analysis of Raska [1983] investigated the corrosiveness of the atmosphere debris samples that were removed from 16 weathering steel along the Texas Gulf Coast by exposing coupons of A588 Grade bridges in Michigan and analyzed for sodium chloride [Culp B steel at bridge sites located in High Island, Corpus Christi, 1980]. Salt concentrations as high as 8 percent were found in and Port Isabel. The exposure racks at High Island and Port the rust removed from flanges beneath leaking expansion joints. Isabel were placed on the bridge piers for the main span about Salt concentrations as high as 3 percent were found where road- 80 ft (24 m) above ground; the rack at Corpus Christi at the way spraying and dusting were the only sources of contami- south end of the US 81 Nueces Bay Causeway about 8 ft (2.4 nation. The saltwater runoff from Michigan's highways can have m) above water. The High Island Bridge is made of weathering a sodium chloride concentration higher than the 3.5 percent steel; the others are not. concentration in seawater. The coupons were tensile tested after weathering. The mea- To determine the section loss of weathering steel bridges due sured loss of load-carrying capacity, as compared to that of the to corrosion, McCrum [1983] measured the thickness of webs nonweathered specimens, was converted into an equivalent uni- and flanges of 52 bridges located in lower Michigan. Twenty- form corrosion penetration. The results are plotted in Figure five bridges were located in rural areas and 27 in urban areas. The corrosion penetration at all three sites exceeded, after Most were less than 8 years old. Comparing the thickness mea- less than one year, the 4 mils (100 m) upper bound of Beth- sured with an ultrasonic gage with the nominal thickness yielded lehem's 18-year scatterband. Salt fog caused the extreme cor- information on corrosion rates and locations of most severe rosion penetration of these weathering steel specimens that were corrosion. The findings were as follows: exposed several miles inland from the coast. Since the specimens were exposed beneath the bridge, on top of the main span pier, Salt, regardless of source, is the major cause of accelerated it appears that the weathering steel superstructure of the High corrosion of weathering steel bridges. Island Bridge is also corroding at an excessive rate. Salt spray from traffic passing under a bridge is as corrosive Hem [1981] vertically mounted Resista HRL 37 and A588 as salt water leaking through expansion joints. The former affects Grade A weathering steel specimens on guardrails along West- large surface areas of the steel beams; the latter is usually limited German highways in Rendsburg, Duisburg, Merklingen, and to areas near joints. Trockau. He also exposed control specimens at reference sites The first three or four beams facing the on-coming traffic near each bridge. The specimens at the highway sites were corrode almost twice as much as those following later in the subjected to deicing salt spray from the roadway, whereas those traffic spray path. at the reference sites were not. The data are plotted in Figure On urban bridges older than 7 years, steel beams nearest The corrosion penetration of the weathering steel specimens to the on-coming traffic beneath the bridge are corroding at a at the four highway sites (curves 1 and 2) were, on average, 1.9, rate of 3 to 4 mils (75 to 100 m)/year/surface on the top of 1.3, 2.1 and 1.9 times that of their counterparts at the reference the lower flange nearest the web and on the lower >'4 of the web. sites (curves 3 and 4).

300 NEWARK, N. J. VERTICAL HORIZONTAL I Interior Girders 2 Fascia Girders 250 3 Root Of N.J. D.O.T.

A Carbon Steel 200 0A242 Type I

E D. / / /

- -A

/ -- 2 'I

Le EXPOSURE TIME, YEAR

Figure 64. Effect of salt contamination on corrosion of carbon and weathering steels exposed vertically and horizontally to industrial atmosphere of Newark, N.J. [Cosaboom 19791

The aforementioned Michigan, Texas, and West-German data lengthens the time of wetness (Chapter Three, "Effect of collected at bridge sites show that the corrosion penetrations Aqueous Environments and Salt"). exceeded Bethlehem Steel's 18-year scatterband of 2 to 4 mils The corrosion penetration at marine sites decreases as the (50 to 100 m) for Mayan R steels after 1 to 3 years exposure. distance from the ocean increases and, correspondingly, as the The corrosion rates did not level off with exposure time. The airborne salt content decreases (Chapter Five, "Corrosion Re- weathering steels were unable to develop a protective oxide sistance"). coating and corroded at the same rate as the carbon steels. Discrete salt deposits can cause pitting (Chapter Five, "Pit- Additional information on the detrimental effect of salt, pre- ting"). sented in other sections, is summarized in the following: When a differential aeration cell is established in crevices, the chloride ions present in the crevice permit the electrochem- When the chloride ions present in salt-bearing water come ical corrosion reaction to continue (Chapter Five, "Crevices"). in contact with the steel, they shift the negative potential of the rusted steel to higher negative values. The ensuing increase in Deicing salt deposits that remain on weathering steel bridge the galvanic potential accelerates metal corrosion (Chapter components for extended periods both inhibit the formation of Three, "Effect of Aqueous Environments and Salt"). a protective oxide coating and break down the passive condition The oxide coating on salt-contaminated steel surface con- that such protective coatings confer. As a result, the corrosion tain large amounts of crystalline nonprotective rust of the form rate is not reduced. Knowing that in the absence of deicing salts /3-FeOOH (Chapter Three, "Effect of Aqueous Environments weathering steels perform effectively, and in the presence of and Salt"). deicing salts the contaminated areas on a bridge structure per- The strong moisture adsorption tendency of salt deposits form unsatisfactorily, the following steps should be taken to 85

EXPOSURE TIME, YEAR

Figure 65. Effect of salt contamination on corrosion of carbon and weathering steels exposed vertically and horizontally to industrial atmosphere of Detroit, Mich. [Zocolla 19761

ensure adequate corrosion performance of A242 and A588 ceptible to salt fog. The area where bridge construction is likely weathering steels: should be carefully examined for the performance of steel structures before a decision is made. Salt-contaminated surfaces should be steam-detergent cleaned at the end of the winter season. It is prudent to spot paint areas which, on the basis of DEBRIS experience, are most seriously attacked by corrosion. The debris that accumulates on the horizontal surfaces of By taking these two forms of maintenance action, one can girder elements consists of road dirt, salt, leaves, rust flakes derive the benefits from the use of weathering steels in those spalling from the webs, and occasionally bird nests and drop- portions of the country where unusually hazardous environ- pings. Long retaining walls along the shoulders and low clear- ments influence adversely the performance of a weathering steel ance over the highway create tunnel-like conditions in an bridge. underpass. These conditions intensify the air blast caused by The environmental hazards are confined to regions of the truck traffic and carry more road dirt and salt spray to the steel country where relatively large amounts of deicing salts are used surfaces of the overhead bridge. Debris accumulates more often and locations where salt fogs and nightly condensation occur. towards the supports and builds up against the web at corners Where deicing salt is used, it is prudent to paint potentially formed with stiffeners. By retaining moisture, debris deposits affected areas. Where salt fogs prevail, it may be prudent to keep chlorides and sulfates in close contact with the steel, greatly paint areas that are not washed by rain. Coastal areas in the prolong the time of wetness, thus inducing, poultice corrosion. northwest and along the Texas Gulf Coast are especially sus- Poultice corrosion feeds on debris from accumulated dirt and 86

their tendency to corrode galvanically in seawater flowing at 7.9-13.1 ft/sec (2.4-4.0 rn/see) for 5-15 days at 41-86 F (5- 30 Q. The farther apart any two metals are in the table, the stronger the metal to the right will corrode. Mild steel (- 0.6 to - 0.7 V) and low-alloy steel (- 0.6 V) have low nobility relative to the other metals and alloys listed in Figure 68. It is possible for certain metals, such as the stainless steels, to reverse their positions in some environments, but the relative positions as given will generally hold in natural waters and in the I Port lsabel,Texas atmosphere. 1170 pm at II Months Berger [1980] reported that, in praetiee, mill scale is niofe 1960 pm at 27 Months noble than steel and is found to be one of the more common 2 Corpus Christi, Texas causes of pitting. Few data quantitatively document the relative 3 High Island, Texas nobility of mill scale. Potential measurements for mill scale in seawater at a velocity of 2 ft/sec (0.6 rn/see) over a period of E 31 135 days showed a fluctuating value of 0 to - 0.1 V against a saturated calomel reference electrode [LaQue 1955]. Comparing this value with the potentials listed in Figure 68 indicates that mill scale is 0.6 V more noble than mild steel and 0.55 V more noble than low-alloy steel. Thus, in saltwater immersion, mill scale can cause galvanic corrosion and pitting at breaks in the mill scale [Coburn 1974]. Under conditions of alternate wetting and drying, the mill scale slowly disappears due to undercutting of the scale in the cracked areas. To avoid deep pitting in locations that are continuously damp, the mill scale should be removed before the particular component of a structure is placed in service. The area ratio of two galvanically coupled metals under immersed conditions affects the degree of corrosion of both metals. Phelps and Schmitt measured the relative corrosion rates of carbon, A242, and stainless steels when tested alone or coupled to each other [Phelps 1970]. The couples were immersed in kre standing seawater for 6 months. The area ratios of the specimens EXPOSURE TIME,YEAR were 1:8, 1:1, and 8:1. The corrosion penetration of A242 steel Figure 66. Effect of salt contamination on corrosion of weath- alone was 4.1 mils (105 rn), and of carbon steel alone 5.0 mils ering steels exposed to the Texas Gulf Coast atmosphere. [Raska (128 m), as shown in Figure 69. By coupling the steels, the 1983] corrosion penetration decreased for the cathodic A242 steel and increased for the anodic carbon steel. Most notably, the corrosion of carbon steel tripled when its area was one-eighth that corrosion products. For this reason, the debris should be pe- of the A242 steel. Low carbon-to-A242 steel area ratios can riodically removed from all steel surfaces, preferably once a occur in practice when bolts and welding rods are less noble year. New designs should avoid confined areas (Chapter Eleven). than the high-strength low-alloy steel. Under and conditions, no harm will occur. However, under persistently damp conditions, the less noble member of the couple will corrode at an GALVANIC CORROSION accelerated rate in the contact areas. Bolts joining weathering steel members should have a com- When two dissimilar metals are in contact in a moist or position similar to the weathering steel, because carbon steel immersed environment, the difference in solution potential be- bolts would corrode at a more rapid rate in an aggressive en- tween the two induces a current flow through the solution vironment. An example of such behavior was found in a bolted (electrolyte) from the less noble (anodic) to the more noble A242 Type 1 steel angle specimen that was weathered for 13 (cathodic) metal and back through the metallic contact. Such years in marine Point Reyes, California [Reed 1982]. In this a galvanic couple causes the anodic metal to corrode more than test, 60 percent of a low-carbon steel nut corroded away. it would when exposed alone in the same environment, and the Similarly, to avoid galvanic corrosion in welds, the rod com- cathodic metal to corrode less than when exposed alone. positions should be selected on the basis of their being noble to The degree of corrosion depends partly on the relative position the base steel. Single pass welds may absorb enough alloying of the two metals to one another in a galvanic series. One such elements from the weathering steel base to preclude galvanic galvanic series is given in Figure 68. It should not be confused corrosion. But the external or final passes of multipass welds with the similar electromotive force series that shows exact should have a composition similar to the steel base, when the potentials based on highly standardized exposure conditions thickness exceeds 0.25 in. which rarely exist in nature. The galvanic series of Figure 68 In the Phelps-Schmitt experiments, when coupled to Type is a list of common metals and alloys arranged according to 410 stainless steel, both carbon steel and A242 steel became the

HiQhway Sites DUISBURG MERKLINGEN TROCKAU I HRL37 Steel

2 A588 Grade A

Reference Sites 3 HRL 37 Steel 4 A588 Grade A

RENDSBURG E I50 2

/J:I2

5 0 5 0 EXPOSURE TIME, YEAR

Figure 67. Effect of salt contamination on corrosion of weathering steel specimens exposed vertically to various atmospheres in FR. G. [Hein 1981]

LI C)- -o ID-I ID

(DC FiUaufl ilDn onn -IC ) -I. LJu - UI 0 90 00 I) - C 0 -. - -. D9 OIDI r r.3 ID <---I. —C)(D 00,0000000 C - (.1)0. ID -I 0 - C) ID ID ID C) C) C) - > 0 C) - I II) -IID00 0 1 ID V10 N) W >0 C) C) I*(D C)) -0 - 0ID1 ID ID-I II 1 CO C) C) 0 ID C ID 1 ) < N C)IDID 00UI IDIU) IDID - - ID ID U) ID UI •--1* rD I 1* DO -I* )ID'< C II ! UI ID ID > C ID ID CD < ID 0 W ID UI01. - I 1. ))(D -1 WID a 1*0 LI UI'< I N)1) ID >U)1* UI 0 C) ID 0 ID UI -1. C) (D UI ID '<1•lID ID - -.. ID UI) ID 1* Un UI Q 0 .- UI C) C) UI Q. N UI U 0 0.0. 0 ID C) UI UI UI ID UI CD C) 1. N ID ID - - U) ID CD 1< 1* I 1* ID 1* 1* C) ID = 90 0 IID CD ID C-.- - l.UIID ID M . ID 0 01*- UIID -4ICD< UI- - - -- S > I-P 0 C) I I I CD7r10- CD - ID Ni (D C) 03 -1-IC) C).... ID UI 1 0 1 -I ID ID ID < 1< 1< 1< < ID C) U' -C) 0. (1) -o 0 UI ID pP - — ID 01 CD (D 50) UI C) UI ID • C) -ID U) -) CD ID 0(.1)- (1) .ID UI 0 > C (1) N) C) C) l- ID 1 ID 0 UI 03 '< C 1* -, UI ID -IDC) — ID 0) UI - UI D

(ACTIVE) Figure 68. Galvanic series of various metals in flowing seawater flowing at 2.4-4.0 rn/s for 5- 15 days at 5-30 C. [Clark 1978] 88

In all instances, the chloride ions from deicing salt or seawater E spray are the primary stimulants that accelerate pitting. Dif- z ferential aeration cells, resulting from scattered deposits of damp 0 dirt, leaves, twigs and oil, likewise stimulate pitting of the surface .— of all steels, from carbon through stainless. In view of the fore- Cr going, it is evident that: I— wz w Bridges covered with mill scale may develop numerous 0 z pits at cracks in the scale when these cracks are covered with 0 dirt, leaves, etc. and are damp for relatively long periods of time. Bridges that are free of mill scale which, because of their Cr 0 geometry, trap and retain drainage, debris, windblown dust, U Alone 0.125 1.0 8.0 Alone dirt, deicing salts and maintain a damp condition can develop deep pits beneath the deposits.

CARBON/A242 STEELS AREA RATIO As part of the corrosion study cited in Chapter Five, under Figure 69. Corrosion penetration of uncoupled and coupled steels "Industrial Environments" and "Marine Environments," Cop- after 6 months in seawater. [Phelps 19691 son measured the average depth of the four deepest pits on the skyward and groundward surfaces of many steel specimens [Copson 1960]. Figure 70 shows the pitting and uniform corrosion of steels with chemical compositions typical of copper- bearing and weathering steels, as a function of time of exposure sacrificially corroding (anodic) member of their respective cou- in industrial Bayonne, N.J., and marine Block Island, R.I. Pit- ples, regardless of area ratio. ting and uniform corrosion at the marine site were more severe The galvanic couple created by the bronze washers in the than at the industrial site. For all steels tested by Copson, pit expansion joints of some Michigan bridges contributed to the depths after 17 years were two to three times the average pen- corrosion of the web and link plates [Cuip 1980]. The main etration calculated from weightS loss. The pitting factor (ratio factor, as discussed under "Crevices," was crevice corrosion. of pit depths to average corrosion penetration) for the low-alloy The severe corrosion, including as much as 0.25 in. (6 mm) steels plotted in Figure 70 was about three. Pit depths tended deep pitting, restricted the movement of the expansion joint and to increase with time in a manner similar to that of average required replacement of the link plates and pins. corrosion penetration. The phenomenon of galvanic coupling is also relevant to The Copson data and the experience with the Michigan crevice corrosion, as discussed under "Crevices," later in this bridges and the Doullut Canal Bridge on the Mississippi Delta, chapter. Louisiana, indicate that salt contamination stimulates conditions responsible for both pitting and general corrosion. [Copson 1960, Culp 1980, Arnold 1981, Gallagher 1982]. PITTING The effect of pitting on fatigue life is discussed in Chapter Eight, under "Application to Current Designs." Pitting is a complex corrosion phenomenon. It is the initial form in which all forms of corrosion manifest themselves. If one visualizes the surface of a carbon steel structure to be in the form of a grid, such as a sheet of graph paper with an infinite CREVICES number of small containers filled with a saltwater electrolyte, then corrosion begins simultaneously in random locations in Crevice or joint corrosion is a form of localized attack of a numerous but not all of the minute containers. The steel in the metal surface at, or immediately adjacent to, an area shielded adjacent containers will not corrode for some time because they from bold exposure to the environment. Bolted connections, are galvanically protected by being coupled to the adjacent cantilever expansion joints, and lap joints not welded all around corroding areas. The result is called general corrosion. The are examples of details susceptible to crevice corrosion. identical action occurs with the weathering steels. It is stimu- The mechanism of crevice corrosion involves electrochemical lated by the physical and chemical heterogeneities present on reactions that take place between the surfaces within the crevice, the surface. or between the surfaces of the crevice and those freely exposed The pitting is generally of a shallow nature, with the weath- to the environment outside the crevice [Ellis 1951]. Crevice ering steels showing the least signs of deep pitting. Mill scale corrosion of weathering steels occurs in two stages. At the initial tends to generate somewhat deeper pits along cracks in the scale, stage, when the protective oxide coating has not yet formed on because the relatively more noble potential of the scale, com- freely exposed surfaces, a differential aeration cell is established pared to the base steel, forms a strong galvanic cell. This con- between the steel within the crevice, which is shielded from dition is not too serious since it provokes additional fracturing access to the dissolved oxygen in the electrolyte, and the steel and eventual spalling of the mill scale. As an example of another outside the crevice, which is freely exposed to the oxygen-bearing metal, pits form in stainless steels much less frequently. How- electrolyte. Lacking a supply of oxygen, the steel inside the ever, the individual pit can perforate the steel in a relatively crevice becomes anodic and corrodes. short time. After several years exposure, the protective oxide film covers 89

I Pitting Corrosion I 2 Uniform Corrosion / BAYONNE. N.J BLOCK ISLAND, R. I. - - - Copper Steel Low Alloy Steel / I

Le 20

Figure 70. Comparison of pitting and uniform corrosion of copper steel and low-alloy steel exposed in industrial Bayonne, N.J., and marine Block Island, R.I. [Copson 1960]

the freely exposed surfaces, but not the steel within the crevice. joint and caused other structural damage to the bridge. After At this stage, in addition to the differential aeration cell, a sandblasting, it became apparent that the web behind the link galvanic cell is established between the active steel within the plate was more severely corroded in an area around the bronze crevice and the passive steel outside the crevice. The difference washer outline than in contact with the washer. This points to in potential between the active and passive states of the steel crevice corrosion as the dominant corrosion mechanism [Fig. becomes an additional driving force for galvanic corrosion of 13, Culp 1980]. the steel within the crevice. Among the oldest weathering steel structures in Michigan is In the presence of salts and to a lesser degree various corrosion a test stretch of guardrail installed in February 1963 along a products, the differential aeration cell electrochemically trans- highway in Lansing, Michigan [McCrum 1980]. An examination ports the salts' anions, particularly chlorides, into the crevice. of the guardrail in 1978 revealed bulging lapped joints expanding Simultaneously, anodically dissolved iron ions and salt anions outward from the internal pressure of the corrosion products. form acidic salts. This, in turn, lowers the pH value of the This condition has also been termed "packout." Ultrasonic electrolyte within the crevice. The combination of high concen- thickness measurements indicated a 40 percent section loss of trations of anions, low pH, and low oxygenconcentration causes the guardrail in the lapped areas. The average effective corrosion high corrosion rates and promotes pitting. rate was 1.4 mils/year/surface (35 sm/year/surface). That is The following two cases illustrate the deleterious effect of almost three times the corrosion rate of the freely exposed guard- deicing salts on corrosion of steel within crevices. Visual in- rail surfaces. Some areas of several square inches exhibited six spection of a disassembled link plate in an expansion joint re- to seven times the corrosion rate of the freely exposed surfaces, moved from a weathering steel bridge in Detroit, Michigan, or about 3 mils/year/side (75 j.m/year/side). revealed severe crevice corrosion between the link plate and the Adequately tightened and stiff high-strength bolted joints web. It resulted in tight rust packing of the gap between the should not normally be prone to crevice corrosion, as is discussed two elements [Culp 1980], prevented movement of the expansion in Chapter Ten. 90

CHAPTER SIX

STRENGTHENING MECHANISMS AND TOUGHNESS

INTRODUCTION point is a linear function of d'2, where d is the mean grain diameter [Petch 1959]. Weathering steels for bridges are supplied to ASTM Speci- Ferritic-pearlitic low-carbon steels, such as the A588 steels, fications A242, A588, and A709 which set minimum require- can be grain refined by: (1) rolling at reduced temperature, or ments for tensile properties, that is yield point, tensile strength, controlled rolling; (2) alloying; and (3) normalizing. and elongation. Controlled rolling involves large thickness reductions at the Different manufacturers have developed their own composi- lower hot-rolling temperatures and may involve accelerated tions for meeting the specification requirements, and this is cooling rates. This produces fine austenite grains that do not reflected in the different grades that are available in A588 and have sufficient time to recrystallize at the low temperatures in the different compositions actually used to meet A242 and before austenite transforms to ferrite upon cooling. The resulting A709. The weathering HSLA steels A242 and A588 generally ferrite grains are then relatively fine. Grain size is increasingly have a minimum yield point of about 50 ksi (345 MPa) and are refined as rolling temperature is lowered and reduction at the usually supplied in the as-hot-rolled condition. The weathering lower temperatures is increased while still staying above the alloy steels A709 Grade 100W have a minimum yield strength transformation temperature. Only light-gage product is usually of 100 ksi (690 MPa). They are quenched and tempered alloy control-rolled. steels, essentially the same as the A514 steel with its numerous Certain alloying elements help to refine grain size. Their main grades. function is to form precipitating compounds in austenite (alu- All of the above-named steels are of structural quality. They minum), and to retard the growth of recrystallized austenite are frequently not supplied to any notch toughness requirements. grains during further hot rolling (aluminum, manganese, chro- On Federal-aid highway programs in the United States, how- mium, molybdenum and nickel). Maintaining a fine austenite ever, Charpy V-notch toughness requirements have been grain size results in fine-grain ferrite on subsequent transfor- adopted for primary tension-loaded members of bridges, as listed mation. in specification A709. Thus, when agreed on by the manufac- Normalizing is a process in which an alloy steel is heated to turer and the purchaser, CVN impact tests may be included for a suitable temperature above the transformation temperature bridge steels as a supplementary requirement, with the tests range and subsequently cooled in still air at room temperature. conducted in accordance with the requirements of specification The presence of aluminum, columbium, or vanadium inhibits A673. coarsening of the austenite grains of the reheated steel and results The imposition of notch toughness requirements may well in fine pearlite formation and grain refinement. This, in turn, necessitate making some of the steels with slightly different promotes toughness and lowers the impact transition temper- compositions and deoxidation practices as well as the addition ature. Normalizing only refines the grain if the proper grain- of heat-treatment. refining elements are present. To assess the performance of weathering steel, the designer The A588 steels are all made by fine-grain practice but are should have some knowledge of the strengthening mechanisms usually supplied as rolled. Even on fine-grain-practice steels like that the suppliers use. Fletcher and Porter, in reviewing the A588, the grain size of hot-rolled product will be much the evolution and physical metallurgy of high-strength low-alloy same as that of semikilled steel. The fine-grain practice itself, steels, discussed the various strengthening mechanisms com- however, does impart better notch toughness. monly used in practice [Fletcher 1979; Porter 1982]. The mech- In hot-rolled steels, the grain refinement method of strength- anisms that apply to weathering steels are summarized in the ening is unique in that it concurrently improves impact prop- following sections. erties. For example, Pickering and Gladman found that the yield and tensile strength of hot-rolled, low-carbon steels with ferritic- pearlitic structure increased and their fracture-appearance transition temperature decreased with grain size. The relationship STRENGTHENING MECHANISMS between the aforementioned mechanical properties and d 2, is linear [Pickering 1963]. High strength in bridge steels can be achieved with several Solid-Solution Strengthening. Solid solution of any alloying strengthening mechanisms. Steel manufacturers may choose a element in a metal strengthens the solvent metal. The differences combination of these mechanisms to produce a weathering steel in atom size and electronic structure between the solute alloying with desirable yield and tensile strength as well as toughness. element and the solvent metal determines how much strength- Grain Refinement. One way by which steels can be strengthening is obtained. In dilute solid solutions with up to 1 percent ened is to decrease the grain size; in other words, to increase of any common solute and up to 1.8 percent manganese, the grain-boundary area per unit volume. Indeed, the lower yield amount of strengthening is about proportional to the concen- 91 tration of the alloying element. If more than one solute element to ensure good corrosion resistance, and minimum specified is present, the total strengthening is about the sum of the effects lower yield point and tensile strength. More often than not, the of each alloying element separately. The alloying elements of upper limits provide simply suitable melting ranges for the ele- steels are either small enough to fit between the iron atoms in ments or help to provide the specified strength in the heavier the crystal lattice, or they are of a size so that they can substitute thicknesses. Transition temperature is sometimes a consideration for some of the iron atoms. They are termed interstitial and in setting upper limits on alloying element content. substitutional alloying elements, respectively. Carbon and ni- Precipitation Strengthening. Reactions leading to precipitation trogen are examples of interstitial alloying elements. Other solid- strengthening markedly improve the lower yield point of the solution elements, such as phosphorus, silicon, copper, man- steel beyond what might be obtained by solid-solution strength- ganese, molybdenum, and nickel are substitutional. Vanadium ening. and columbium, on the other hand, are not solid-solution ele- A number of alloying elements precipitate from supersatu- ments. They strengthen the steel mainly by precipitation. rated solid solution upon cooling. The microalloying elements, To achieve a high-degree of solid-solution strengthening, a vanadium, columbium, titanium, and aluminum precipitate as large difference between the atomic sizes of the solute and the nitrides, carbides, and carbonitrides. Copper precipitates by it- solvent is required, so that the dislocations of the solvent metal self. The degree of strengthening depends primarily on volume are locked by its solute atmosphere. For this reason, elements fraction and also on the size and spacing of the precipitates. like carbon, nitrogen, and phosphorus, which are relatively small Aging time and temperature markedly affect the size and spacing compared with iron, increase the strength most, whereas nickel of the precipitate particles, because they relate to the degree of and chromium, which are comparable in size with iron, do not supersaturation and the volume fraction of the precipitate par- increase the strength. Figure 71 shows the effect of common ticles. The necessary condition for precipitation strengthening alloying elements on the lower yield point and the tensile in an alloy is a sloping salvus line in the phase diagram, so that strength of weathering steels. Apparently, chromium weakens there is appreciable solubiity of the precipating phase at elevated the steel because it removes some carbon and nitrogen from temperatures and a marked reduction in solubility at lower solid solution. temperatures. Columbium and vanadium are the two most useful Solid-solution strengthening can be detrimental to impact alloying elements for this purpose. They are used for strength- properties. Additions of most alloying elements, especially in- ening many HSLA steels. terstitials as well as phosphorus, raise the impact transition Vanadium causes marked precipitation strengthening and temperature. Manganese increases strength as well as notch usually some grain refinement on normalizing, even in thick toughness and is inexpensive. For this reason, just about all the plates, because all of the vanadium is in solution at usual nor- HSLA steels contain a high manganese content. Nickel improves malizing temperatures. In concentrations up to 0.10 percent, toughness, but it has no effect on yield point and only a minor vanadium seems to have negligible deleterious effect on tough- effect on tensile strength. ness. Also, in semikilled steels, vanadium combines with part The A588 and A709 specifications set both lower and upper of the nitrogen, thereby reducing the free nitrogen content and limits on alloying element content. The lower limits are meant further lowering the impact transition temperature.

ALLOYING ELEMENT,% Figure 71. Effect of solid solution alloying elements on change in (a) lower yield point, and (b) ultimate tensile strength. [Pickering 19761 92

Columbium provides marked precipitation strengthening in bon content of HSLA steels is generally kept below some max- as-rolled steels. Its use in as-rolled steels, however, is limited to imum value that depends on the overall chemical composition % in. (19-mm) thick product (ASTM Specificaion A572), unless and the intended application. The carbon content of A588 steels the steel is killed. In this case, there is no specified maximum suitable for use in welded bridges is limited to about 0.20 percent thickness beyond those of the A572 specification. maximum. When the steel is normalized at the usual austenizing tem- Chromium. Chromium can potentially increase strength by perature, columbium refines the grain, thereby improving tough- solid solution and by reducing the interlamellar spacing of pear- ness and lowering the impact-transition temperature. In this lite. On the other hand, it removes carbon and nitrogen from case, columbium does not appreciably strengthen the steel solid solution. The net effect is that chromium weakens the through precipitation. steel, as shown in Figure 71. To improve both strength and toughness of columbium-bear- Columbium. Columbium increases the yield point and the ing steels, controlled-rolling would be required. Indeed, some tensile strength of as-rolled steels by precipitation strengthening weathering steels may be control-rolled to meet AASHTO notch and, to a lesser degree, by solid solution. It also refines the grain toughness requirements. size when the steel is control-rolled. Another advantage of al- Quenching and Tempering. Quenching and tempering is a loying with columbium is the element's low affinity for oxygen, heat treatment of steel that increases strength, hardness, and which maximizes the production yield of semikilled steels from toughness by virtue of rapid cooling (quenching) from above the ingot. The impact transition temperature of conventionally the transformation range. Quenching produces martensitic or rolled HSLA steels containing columbium is high, because most lower bainitic microstructures which is more desirable for of the strength is achieved through precipitation. Hence, if low strength and toughness than the ferritic-pearlitic microstructure. impact transition temperature is required, steels containing co- After quenching, the steel is in a highly stressed condition. lumbium are usually normalized or control-rolled. To avoid or minimize problems of cracking or distortion, the Copper. Copper increases the yield point and the tensile steel is then heated to an elevated temperature and air cooled, strength primarily through solid solution. The effect of copper a process called tempering. The best combination of strength as a solid-solution strengthener is shown in Figure 71. To achieve and toughness is usually obtained by suitably tempering a precipitation strengthening, the steel has to be alloyed with at quenched microstructure consisting of a minimum of 80 percent least 1.3 percent copper and also age hardened. Weathering martensite throughout the cross-section. steels contain much less copper and are, therefore, not strength- The A709 Grade 100W weathering steels are quenched and ened by precipitation. Weldability and toughness are not affected tempered. by copper contents below 0.75 percent. Steels containing more than 0.50 percent copper frequently exhibit "hot shortness" during hot working, so that cracks or extremely roughened ALLOYING ELEMENTS surfaces may develop during hot working. Adding nickel alle- viates this problem. Copper, especially in amounts up to 0.25 Weathering steels may contain up to 15 different alloying percent, is the element most beneficial in improving atmospheric elements in various combinations and proportions. It was pre- corrosion resistance. viously noted that the presence of some of these elements in Manganese. The chemical interaction of manganese with sul- natural iron ore (i.e., sulfur) or their addition to steel (i.e., fur and oxygen makes it possible to roll hot steel. Manganese copper) may influence strength and toughness in addition to increases the yield point and tensile strength by substitutional atmospheric corrosion resistance. Some elements are used be- solid solution, as shown in Figure 71, and improves toughness. cause of the specific properties they impart to steel. Others rid In addition, it is a strong hardener, which is important to the steel of impurities or render the impurities harmless. A third quenched and tempered steels. Higher ratios of manganese to group counteracts harmful oxides or gases in the steel. The carbon content prevent the formation of acicular transformation elements of this latter group do not remain in the steel to any products, which are harmful to ductility and toughness. Fur- great extent after it solidifies. Some elements fall into more than thermore, by refining the grain size, manganese compensates in one of the aforementioned groups. The following discussion part for the loss of ductility and toughness resulting from the describes the effect of the more common alloying elements on addition of other alloying elements. To ensure good weldability, strength and toughness. the manganese content should be kept below some maximum Aluminum. Aluminum is generally used as a deoxidizer; usu- value that depends mainly on the content of carbon and other ally along with silicon, when fme-grain-practice killed steel is alloying elements. This is also discussed in Chapter Ten. The specified. It does not have much, if any, effect on grain size in lower limits of manganese content specified for the various the as-rolled condition, but it does improve notch toughness. grades of A588 steel relate primarily to strength and notch When such is normalized, however, the grain size decreases toughness, the upper limits often relate to weldability. substantially and the transition temperature is lowered. Alu- Molybdenum. Molybdenum is an effective substitutional minum in concentrations greater than 0.1 percent is a solid- solid-solution strengthener. However, it reduces the effectiveness solution strengthener. However, because steels containing alu- of columbium as a precipitation strengthener. High molybdenum minum tend to have poor surfaces, this element is not used for contents harden the heat affected zone (HAZ), which is a dis- solid-solution strengthening. advantage in welded steel construction. In as-rolled steel and Carbon. The simplest and cheapest way to increase the tensile possibly in normalized steel, molybdenum tends to impair tough- strength of steels is to raise the carbon content so as to provide ness, even when present in relatively small amounts. It is a a greater percentage of pearlite in the microstructure. The ad- nonoxidizing element, making it highly useful in the melting of dition of carbon does not markedly increase the yield point, and steels, such as quenched and tempered steels, whose hardness impairs ductility, toughness, and weldability. Therefore, the car- must be closely controlled. 93

Nitrogen. Free nitrogen, that is nitrogen in interstitial solid TOUGHNESS solution and not combined as a nitride, strongly strengthens the steel. But, it has the disadvantage of increasing the impact Toughness of a steel can be defined as the ability to absorb transition temperature by 50 F (28 C) for each 2.0 ksi (14 MPa) energy by undergoing plastic deformation prior to fracture. The increase in yield strength [Fletcher 1979]. measure of this ability at a high rate of loading is called impact Nickel. Nickel marginally increases the tensile strength and strength. Notches or stress concentrations are present in nearly has no effect on the yield point, as shown in Figure 71. Sub- all structures. If a steel is to resist fracture under impact loading, stantial amounts of nickel effectively lower the impact transition it must be able to absorb considerable energy or show notch temperature. Nickel in an amount equal to at least one-hall the toughness whenever stress concentrations occur. A component copper content is very beneficial in preventing the hot shortness can fracture at points of stress concentration, even though the of steels containing about 0.50 percent or more copper. nominal stress is below design values, when the steel is subjected Phosphorus. Phosphorus is a strong substitutional solid so- to static or impact loads at comparatively low temperatures. lution strengthener (see Fig. 71). Because it markedly increases Body-centered-cubic (BCC) or ferritic alloys exhibit a signif- the impact transition temperature, weathering steels used for icant change in toughness over a range of temperatures. Figure bridges have a maximum phosphorus content of 0.04 percent. 72 shows qualitatively the effect of temperature on toughness. The mill scale on a copper-phosphorus steel is shed more quickly At elevated temperatures, notched specimens fracture by a shear than on carbon steel [Chandler 1970]. mechanism absorbing large amounts of energy. This is associated Sulfur. Although sulfur markedly strengthens the steel by with a ductile, fibrous appearing fracture with plastic defor- substitutional solid solution, it is very detrimental to toughness. mation. At low temperatures, notched specimens fracture by a Elongated stringers or ribbons of manganese sulphide, especially cleavage mechanism absorbing little energy. This is associated in more or less coplanar aggregates, cause anisotropy with regard with a brittle, granular-appearing fracture with little or no plastic to ductility and toughness. Reduction of sulfur content to very deformation. At intermediate temperatures, steels exhibit a duc- low values and sulfide inclusion-shape control helps to alleviate tile-to-brittle fracture transition. this amsotropy. With inclusion-shape control, a process that Fracture toughness testing to ASTM E399 specification is reduces the plasticity of inclusions by addition of titanium, very time consuming and costly. For this reason, the impact zirconium, cerium or rare earth mixtures, or calcium, the in- Charpy V-notch test is used as an inexpensive and rapid way clusions remain globular or equiaxial during hot-rolling. Re- of characterizing the toughness of structural steels. Fortunately, duction of sulfur content lowers the impact transition because the curves for CVN energy-absorption and plane-strain temperature. Adding elements that globularize the sulfides, such fracture toughness, Kid, have a transition in the same temper- as rare-earth metals or calcium, raises the Charpy V-notch shelf ature zone, the value of the latter in the transition region can energy, particularly in the transverse direction where the CVN be estimated from [AASHTO 1978]: shelf energy may be low due to straightaway rolling. This elevation is reflected along the whole CVN curve, but to a lesser extent at the bottom of the curve. 5 (CVN) Silicon. Silicon is commonly added to steel as a deoxidizer to produce killed steel. It also increases the yield point and AASHTO has minimum CVN toughness requirements for tensile strength by substitutional solid solution. Silicon in fracture critical members at the minimum service temperatures amounts up to 0.30 percent lowers .the impact transition tem- corresponding to Zones 1 to 3. Table 37 summarizes these perature. Higher amounts reduce ductility and weldabiity. The requirements for A588 steels. The, requirements for AS 14 steel upper limit on the range of silicon content in A588 steel varies should also apply to A709 Grade 100 W weathering steel. from 0.30 percent to 0.90 percent. There appears to be a trade- The toughness and temperature at which the transition from off in the use of silicon as an economically attractive strengthener ductile to brittle fracture occurs can greatly vary from one steel and its harmful effect on toughness, ductility, and weldabiity. to another. Good impact strength is promoted by: low carbon, Titanium. In the past, titanium was added to HSLA steels phosphorus, and sulfur contents; high manganese-to-carbon ra- primarily to refine the grain size and to improve sulfide mor- tio; small grain size; reduced section thickness; normalizing; and phology. More recently, it is being added to prevent grain coarsening of the HAZ and, hence, to ensure good toughness. Titanium reacts with oxygen, nitrogen, carbon, and sulfur in the steel, forming a number of compounds, including oxides, mtrides, carbides, and sulfides. Some of these compounds have

desirable effects, but the presence of others is undesirable. The I Transition Ductil e formation of titanium oxide is undesirable because it entails a Zone LSheI f Energy loss of titanium which then is unavailable for grain refinement, o 03 precipitation strengthening, or shape control of sulfide inclu- 0 0 sions. Hence, the molten steel must be thoroughly deoxidized I)- before titanium is added. 00 Vanadium. Vanadium is an effective precipitation strength- ow ,iTronsIon Temprature ener. In contrast to columbium, it forms precipitation phases Lu Brittle in both as-rolled and normalized steels. A content of 0.08 percent vanadium strengthens steel about as much as 0.02 percent co- TEMPE RATURE lumbium. The combination of columbium and vanadium gives a higher yield strength in the conventionally hot-rolled condition Figure 72. Effect of temperature on toughness or Charpy V-notch than can be obtained with either element alone. energy absorbed.

Table 37. Base metal Charpy V-notch requirements for fracture critical members of A588 steel. [AASHTO 1978]

Zone 1 Zone 2 Zone 3 Thickness -17.8C -18.3C to -34.4C -35C to -51.1C (0.F) (-1F to -30F) (-31'F to -60F)

Up to 102 inn (4 in.) 33.9 J at 21.1C 33.9 J at 44C 33.9 J at -122.2'C mechanically fastened (25 ft-lb at 70'0 (25 ft-lb at 40F) 25 ft-lb at 10SF) or up to 51 ml (2 in.) welded

Over 51 m (2 in.) to 40.7 J at 21.1% 40.7 J at 4.4% 40.7 J at -12.2'C 102 inn (4 in.) welded (30 ft-lb at 70F) (30 ft-lb at 40F) (30 ft-lb at 1O'f)

quenching and tempering. It is, therefore, possible to equate a TEST TEMPRATURE (SF) given type and grade of steel with a range of toughness, but not -tOO -50 0 50 100 150 I I U p ' F with a specific value of toughness. As an example, Figure 73 :40 180 shows CVN toughness versus temperature curves for specimens I 0 1k' Flange 0 cut from A588 steel plates of various thicknesses. It illustrates W36x 230 2A2"Plate the large differences in transition temperature that may occur, 00 3 V 3" Plate a ISO and the need for measuring CVN toughness, to ensure that the 4 0 3/8" Plate steel is suitable for the intended application. 5 0 3/4' Web W36x 230 0 2 Figures 74 and 75 summarize the plane-strain fracture tough- 60 20 ness data for 1-sec loading, K,, and dynamic (about 1 msec) 0 3 loading, Kid, that Crosley and Rippling are collecting for an on- a going FHWA study on fracture of bridge steels [Crosley 1983]. 120 90 The data in Figures 74 and 75 are for specimens that met the B A ASTM E399 size requirements for plane-strain fracture tough- 00 V 4 ness. Additional data for specimens that had higher toughness 80 60 and did not meet the size requirements were not plotted. The 6 o A 1-sec rate of loading is typical of that occurring in bridges. A 0 preliminary analysis of all data collected by Crosley and Rip- 40 30 pling appears to confirm that a correlation exists between the A A Charpy and dynamic fracture toughness measured at the tran- A sition temperature. The transition for 1-sec loading occurs at 0 I I -75 -50 -25 0 25 50 75 lower temperatures. Roberts, et al., reported that A588 steels exhibit large vari- TEST TEMPRATURE, C ability in fracture toughness and a marginal temperature shift Figure 73. Charpy V-notch toughness data for A588 steels. [Rob- [Roberts 1974, Roberts 1977]. With regard to the fracture re- erts 1977] sponse of a welded bridge member, they found that the largest tolerable crack at the lowest service temperature, as provided for by the AASHTO toughness requirements, will generally be separation of the inclusions. Those properties involve ductile of the order of the flange plate thickness. Thus, one can conclude failure in which the nonmetallic inclusions nucleate voids that from these results that the specifications provide welded details grow and coalesce to form a major fracture. Elongated stringers that will nearly exhaust all fatigue life before a fatigue crack of manganese sulfide, MnS, or discontinuous stringers of alu- will become unstable at the lowest anticipated service temper- minum oxide inclusions (Al2O3) cause premature ductile frac- ature. This conclusion appears to be valid for bridge steels in ture. Manganese sulfide is plastic at hot-rolling temperatures, general, excluding steels that are electro-slag welded. and the particles tend to become elongated during rolling. The In recent years, the ductility and the Charpy shelf energy highly elongated inclusions, and particularly the more or less values of HSLA steels in the transverse direction, and partic- coplanar arrays, cause a marked anisotropy of the total ductility ularly in the through-thickness direction, have received much at fracture and of the Charpy shelf energy. Testing in the lon- attention. This resulted in part from lamellar tearing encoun- gitudinal direction (parallel to the rolling direction) gives much tered in welding and cracking or splitting (delamination) that higher energy-absorption values than does testing in the trans- accompanied certain bending operations in hot-rolled, flat-rolled verse or through-thickness directions. This anisotropy is largely products [Takeshi 1975, Heroe 1975]. The total ductility at eliminated if the inclusions are present as small, isolated, non- fracture and the Charpy V-notch shelf energy of HSLA steels deformed particles. Because most commercial steels have a much are reduced markedly by an increase in the volume fraction of greater volume fraction of sulfides (MnS) than oxides (Al2 03), nonmetallic inclusions, by greater elongation, and by decreasing sulfides are the major problem. It can be alleviated by reducing

TEMPERATURE (°F)

60 •• IC I ; S • • S 000 S • S •.• S S • 40 . . . S .

S S S

I I I 50 -125 - 100 -75 -50 -25 0

TEMPERATURE (°C)

Figure 74. Fracture toughness of A588 steels under one-second loading. [Crosley 19831

TEMPERATURE (°F)

-duo -tu -iuv I I I IE ) 80

60 S. S. • ' . : • ! :• S • . .. . • •• • S • - 2C

I I 0 -150 -125 -100 -75 -50 -25 0 TEMPERATURE (°C) Figure 75. Fracture toughness of A588 steels under dynamic loading. [Crosley 19831 96 the sulfur content to very low values and controffing the inclu- TEST TEMPERATURE, (F) sion shape by calcium treatment or by adding microalloying -180 -150 -100 -50 0 50 80 elements such as rare metals, zirconium, and titanium [DuMond 320i I I 1 1240 1977, Fletcher 1979, Wilson 1982]. Figure 76 compares the Charpy V-notch energy of conven- Non Co. Treated 280 210 tional (not calcium-treated) and calcium-treated A588 steels I LS rolled alike. The specimens were oriented longitudinally. The 2 LT C(L Treated notch extended in the two directions of interest for bridge mem- 240 3 LS 180 bers, simulating a crack through the thickness (LS) and across 4 LT the width (LT) of an element. Table 38 summarizes the Charpy V-notch upper shelf energy of conventionally rolled and calcium 200 ISO treated A588 steels. The steel without the calcium treatment exhibited poor isotropy in the three major directions, whereas more isotropy was achieved through reduction and balling of (60 120 the sulfides by calcium treatment. -- Because ductility and toughness usually decrease as strength I I / / increases, these properties should be carefully optimized with / 120 / due regard to the intended application. ASTM does not specify the minimum impact strength of A242 and A588 steels. The Charpy V-notch impact test and the drop 80 60 weight test are listed in the A242 and A588 specifications as U standardized supplementary requirements for use at the option / of the purchaser. ASTM specification A709 specifies the Charpy J / 30 V-notch impact test as a supplementary requirement to be con- 7H ducted in accordance with ASTM A673. Impact test requirements are in relationship to the lowest ambient temperature 0 — I I I expected for the location where the steel is to be used, and are -120 -95 -70 -45 -20 5 30 only applicable when specified in the contract and order. TEST TEMPERATURE(C) Like any as-rolled steel, the notch toughness of as-rolled weathering steel can vary to a considerable degree depending Figure 76. Charpy V-notch energy absorbed by non-calcium- on such features as individual cast or heat chemistry, rolling treated and calcium-treated A588 steels. [Wilson 19821 temperature and technique, product size, and the particular producing facilities involved. Therefore, when notch toughness is a matter of concern, desired minimum values must be incor- porated as part of any inquiry or order.

Table 38. Charpy V-notch upper shelf energy of conventional and calcium-treated A588 steels. [Wilson 19821

Charpy V-Notch Upper Shelf Energy Specimen V-Notch [Joules (ft-lb)] Orientation Orientation Non-Calcium Calcium Treated Treated Steel Steel

Longitudinal Through thickness 313 (231) 290 (214)

Longitudinal Transverse 160 (118) 279 (206)

Transverse Through thickness 81 ( 60) 195 (144)

Transverse Longitudinal 66 ( 49) 206 (152)

Through thickness Transverse 30 ( 22) 140 (103)

Through thickness Longitudinal 23 ( 17) 149 (110) 97

CHAFFER SEVEN

WEATHERING FATIGUE

This chapter examines all data reported in the literature for The AASHTO allowable S-N lines, located two standard specimens that were weathered outdoors for a period of time deviations to the left of the mean, are given by: and then fatigue tested to failure, a condition termed herein weathering fatigue. The uniform basis for comparing the data log Nd = (b - 2s) - m log F,, (30) is defmed first. Thereafter, the fatigue test data for corrosion resistant (weathering) steels and noncorrosion resistant steels in which Nd equals the number of design cycles, F,, equals the are presented. The chapter concludes with a comparison of the allowable stress range, and s equals the standard deviation on data for all steels and an explanation of the observed losses and log of life. The allowable 1,, versus Nd values for redundant load gains in life in terms of pitting and weld-toe rounding. path structures, given in Table 1.7.2Al of the AASHTO fatigue specifications, are rounded coordinates of points that fall on the allowable S-N lines given by Eq. 30. The lines are cut off hor- BASIS FOR COMPARING S-N DATA izontally at the allowable stress range for over 2,000,000 cycles. Since AASHTO has set the allowable S-N lines at two standard The results of the fatigue tests are presented in S-N plots of deviations to the left of the mean, a sample of fatigue test data the base-tO logarithm of stress range, .1 versus the base-tO could be said to meet the fatigue strength of a given category logarithm of number of cycles to failure, N, as shown in Figures if not more than 2.3 percent of the data fell below the allowable 79-90. The features common to the S-N plots and the basis for line. their comparison are summarized in the following. The regression coefficients, b and m, and the standard deviation, s, that were used to calculate the AASHTO allowable 1. The mean regression lines for the sets of data points were S-N lines are summarized in Table 39. calculated by the least-square method using the model: The relative stress range after weathering is the ratio of the stress range after weathering to the stress range before weath- log N = b - m log f, (29) ering, f, 1.1,2. Both are calculated at 500,000 cycles using the applicable regression equations. (See the open circles on the in which b equals the intercept and m equals the slope. All mean lines shown in Figure 77.) The relative loss in stress range runout tests were excluded from the regression analysis. is, therefore:

Table 39. Regression analysis and fatigue notch factors for the data base of current fatigue specifications.

Nusber of Regressionb Stress Range Fatigue Type of Detail Data Points Coefficients Standard at 500,000 Notch Category Tested Included Excluded Intercept Slope Deviation Cycles Factor b m (MPa) Kf

A Rolled bean 28 16 13.785 3.178 0.221 350 1.00 B Welded bean 55 1 13.696 3.372 0.147 235 1.49 C* Transverse stiffener 135 5 12.681 3.097 0.158 180 1.95 C 51-rn attachment 14 0 12.763 3.25 0.0628 149 2.35 0 102-rn attachment 44 8 12.177 3.071 0.108 129 2.72 E Cover plate end 193 0 11.886 3.095 0.1006 100 3.51 Cover plate end:a t > 20 rn (0.8 in) 18 11.849 3.2 0.1943 83.5 4.19

Notes

Calculated as in Ref. [Albrecht 1981]. Intercept C for stress range in MPa.

Reference Point It establishes the severity of the detail before weathering [Al- 500,000 Cycles. brecht 1981] in terms of the factor, K. by which the mean 3SOMPa(50-8 tSl) S-N line for a detail falls below the mean S-N line for Category r,A - A rolled beams. It should be noted that Kf is not referenced to the mean S-N line for polished rotating beam specimens, as is Category A Mon Rolled Beams done in other fields of application. The reference stress range was calculated by substituting in Eq. 29 the values of b and m for rolled beams given in Table 39, as follows:fA = [l013785/ :: Detail X Mean 1 L96 Before Weatheng 500,000] /' = 350 MPa (50.8 ksi). Detail X Mean I After Weathering

WEATHERING STEELS $00,000

CYCLES TO FAILURE, log N Seven studies reported data for weathering steel specimens that were weathered outdoors, at most 11 years prior to fatigue Figure 77. Definition offatigue notch factor and computation of testing [Kunihiro 1972; Albrecht 1980, l982a, and 1983a; Blake loss in stress range and fatigue life. 1982; Yamada 1983b; Barsom 1983; and Fisher 1983a]. The American studies were performed on A588 Grade A and Grade B steels, the Japanese studies on SMA 50. and 58 steels. The LJ,= 1— (31) chemical requirements for these steels are given in Tables 4 and 10, respectively. Since N is an m-th power exponential function off, the The following types of specimens, shown in Figure 78, were tested: base metal [Kunihiro 1972, Blake 1982], groove welds corresponding relative fatigue life after weathering is (f,1 The relative loss in fatigue life is, therefore, given by: ground flush [Kunihiro 1972], groove welds with reinforcement not removed [Blake 1982, Barsom 1983], bead welds (Barsom 1983), transverse stiffeners [Albrecht 1980, 1982a, and 1983a; = 1 GI'L)~ (32) Yamada 1983b], 4-in. (102-mm) long attachments [Albrecht - 1982a], 4-in. (100-mm) long gussets [Yamada 1983b], and cover plates [Fisher 1983a]. Most series of specimens were continu- The losses and N. are sides of the shaded triangle in Figure J.,. ously weathered prior to fatigue testing; a few were alternately 77. This procedure compares experimental data near its center weathered and stress cycled. of gravity, about midway in the range of cycles to failure usually To shorten the length of exposure, Albrecht [1977] tried to measured in testing programs. It estimates losses in life in a artificially weather A588 steel specimens in the laboratory by manner that makes them insensitive to slope variations of the subjecting them to 1,800 wet-dry cycles of 1-hour duration. An two S-N lines being compared. examination of the surface and sections of the oxide coating The results of the regression analyses of all fatigue test with a scanning electron microscope (SEM) revealed, however, data are summarized in Tables 40 and 41, respectively, for that the artifically grown oxide was porous, coarse, and poorly specimens that were fabricated from weathering steels and from bonded to the steel base. On the other hand, the oxide on noncorrosion resistant steels. specimens of the same batch that had been weathered outdoors The calculated losses in stress range and fatigue life due up to 18 months had the hard, dense, and well-bonded structure to weathering are summarized in Tables 43 and 44. The values characteristic of natural exposure. The SEM work confirmed were calculated with Eqs. 31 and 32 using the mean slope, m the need to naturally weather the specimens. = 3.2, of the five regression lines for the Category A to E data For ease of presentation, the fatigue test data are examined (499 points) given in Table 39. The mean slope of the regression in the following by investigator. lines for the data given in Tables 40 and 41(1,572 points) is actually flu = 3.39. Slopes greater than 5.0 were excluded. In this study, a mean slope flu = 3.2 was assumed to make the Base Metal and Groove Welds calculations compatible with the AASHTO data base. When the number of data points in a test series was in- Kunihiro, et al., were the first to study the effect of weathering sufficient to calculate a regression line, the loss in life, N,,, was on the fatigue behavior of bare steel specimens [Kunihiro 1972]. determined as the ratio of the mean cycles to failure of specimens They examined the following variables: (1) base metal and tested at the same stress range. groove-welded ground flush specimens; (2) 0-year, 2-year, and The S-N lines were compared at 500,000 cycles, in the 4-year weathering; and (3) 5 l-ksi (353-MPa) and 65-ksi (451- manner illustrated in Figure 77. The fatigue notch factor is MPa) minimum specified yield point for SMA 50 and 58 steels. defmed as the ratio of the mean stress range for Category A The specimens were 2 in. X % in. X 18 in. (50 mm X 19 mm rolled beams, f,A (solid circle), and the mean stress range of the x 459 mm) long, as shown in Figure 78. They were fabricated detail under consideration, f2 (open circle). and tested by seven participating laboratories. The specimens were fabricated from hot-rolled atmospheric corrosion resistant f,- 350 MPa - 50.8 Ksi steels for welded structures and supplied to Japanese Industrial K1— .1-- . - . (33) J,2 J,2 Standard G3114 [JIS 1981]. 99

Table 40. Regression analysis of fatigue test data for specimens fabricated from weathering steels.

Reference Type of Type of Weatheringa No. of No. of Regression Coeff1cients Standard Steel Detail Time Specimens Specimens Intercept Slope Deviation (Years) Included Excluded b n 1h Kunihiro SMA 50 Base metal 0 54 19 13.8677 3.176 0.3141 2 49 178 15.1199 3.833 0.2167 1972 and 22b SMA 58 4 61 15.2587 3.874 0.2385 b, Kunihiro SMA 50 Groove weld 0 45 17.4371 4.551 0.2860 58 19 16.2795 4.311 0.1915 1972 and ground flush 2 12b SMA 58 4 63 19 15.0050 3.808 0.1752

Albrecht 9588 Manually 0 24 5' 12.4209 3.007 0.1172 15 5 11.6900 2.717 0.1008 1982a Grade A welded transeerse 2 Sc stiffener 2 alt. 12 11.9734 2.797 0.0885 4 15 sc 11.2695 2.529 0.1172 4c 4 alt. 12 11.8110 2.762 0.0737 7c lbrecht A588 102-nix 0 17 13.3987 3.416 0.0939 fr 1982a Grade A attaclonent 2 15 0 13.3419 3.398 0.0905 2 alt. 8 0 11.6714 2.597 0.1962 4 20 0 12.9336 3.234 0.0976 4 alt. 8 0 12.4304 2.982 0.1725

Automatically 0 12 0 13.1897 3.226 0.1692 Albrecht 9588 4c 1980, Grade B welded transverse 3 12 12.4822 3.017 0.1010 stiffener 3 alt. 12 0 12.0186 2.810 0.1022 1983a 6c 8 16 12.7197 3.167 0.1142 3b Blake 8588 Base metal 0 15 19.6049 5.460 0.1556 1982 Grade A as received 3 7 0 17.9190 4.883 0.0405 6 16 0 19.0511 5.336 0.0817

Blake A588 Base metal 0 7 So 24.1233 7.030 0.3860 2b 1982 Grade A blast cleaned 2 6 31.7058 10.070 0.1148 4.5 8 0 18.4416 5.026 0.1003

Blake 958$ Groove weld 0 8 0 17.0948 4.745 0.2034 1982 Grade A 2 6 0 10.5929 2.194 0.0562 4c Yaiiada SMA 50 Transverse 0 6 12.6129 3.025 0.1149 3c 1983b stiffener 2 6 15.4230 4.152 0.1248 4 7c ------4c Ynonada 58$ 50 100-nm, 0 7 11.3791 2.610 0.0405 3c 1983b gusset 2 9 13.9590 3.674 0.1859 4c 4 6 11.4457 2.587 0.1261 Yaaada SMA 50 Transverse 5.5 3 7 -- -_ -- 1983a stiffener

Barsm,n 9588 Groove weld 11-8' 15 0 24.8196 7.466 0.2429 1983 Grade A 11-UI 16 0 20.3126 5.705 0.2621 11-M 12 2b 19.1370 5.234 0.2796 b Barsuvi 8242 Groove weld 11-8 15 l 20.5061 5.774 0.3801 4b 1983 Type 1 11-UI 10 23.7322 7.014 0.3314 1b 11-M 11 28.8012 8.829 0.3698 2b Barsoin 9242 Bead weld 11-R 14 19.7086 5.457 0.3853 1983 Type 1 11-UI 16 27.2324 8.282 0.3599 11-4 13 1 26.9339 8.098 0.2435

Barsoin 9588 Uead weld 11-8 1983 Grade A 11-UI (e) (e) (e) (e) (e) 11-4

Notes

Specimens were continuously weathered, unless noted otherwise.

Runouts were excluded from regression analysis.

When a replicate specimen was a runout, others tested at the sane stress range were also excluded from the regression analysis.

U = rural, UI = urban-industrial, 4 = moderate marine

Fatigue test data could not be obtained from the investigator.

Intercept for stress range in MPa.

The significance of the variables could not be evaluated with the mean of the Category A rolled beams [Fisher 1970] and the statistical analysis techniques, because the experiment had not Category A allowable line. The data suggest the following con- been designed with that purpose in mind. Still, by lumping the clusions: data from the various laboratories and for the two yield strengths, and plotting the results by type of detail, one can 1. Both the 0-year weathered base metal and the 0-year assess the effect of weathering time on fatigue life. weathered groove-welded specimens exhibited the fatigue The data for base metal subjected to 2-year and 4-year weath- strength of Category A, since only 2.7 percent (2 out of 73) and ering are plotted in Figures 79 and 80, respectively. Also shown 1.6 percent (1 out of 64) of the data points, respectively, fell are the means of the nonweathered and weathered specimens, below that line. Their mean lives were slightly longer than the 100

Table 41. Regression analysis of fatigue test data for specimens fabricated from steels with no atmospheric corrosion resistance.

Weathering0 No. of No. of Regression Coefficientsd Standard Reference Type of Type of Time Specimens Specimens Intercept Slope Deviation Steel Detail (Years) Included Excluded b m b Konihiro SM 50 8ase metal 0 44 18 io.gog 2.067 12b 0.4029 1972 and 2 57 14.7022 3.693 0.2523 SM 58 4 62 23 13.1205 3.063 0.3083

Kanihiro SM 50 Groove weld 0 36 14 16.1164 4.071 0.3451 1972 and ground flush 2 56 14.6795 3.689 0.2756 15b SA 58 4 69 19 15.1260 3.893 0.1903 2b Nihei SM 50A Notched plate 0 9 44.8791 14.490 b 0.3662 1978 3 7 4 16.0415 4.420 0.2143 7b Nihei SM 50A Notched plate 0 10 31.4776 10.698 2b 0.2983 A = -1 3 7 27.5187 9.519 0.3804

Nihei SM 508 Groove weld 0 16 2 19.2248 5.454 0.1526 1978 as welded 3 7 3 13.8379 3.389 0.1127 b Nihei SM 50A Groove weld as 0 30 il 15.3963 3.733 0.2373 4b 1978 welded, R = -1 3 7 12.4359 2.754 0.1803 3c Yanada SM 50 Transverse 0 6 13.9156 3.592 0.0853 1983b stiffener 2 3 6' ------4 3 7c ------

Yanada SM 50 100-.me gusset 0 10 0 11.7209 2.733 0.1099 1983b 4 9 0 11.2160 2.487 0.1358

Notes:

All specimens werecontinoously weathered. Ronoots were eocladed from regression analysis. When a replicate specimen was a runvot, others tested at the sane stress range were also excluded from the regression analysis. Intercept for stress range in MPa.

4WL~~~Jm* / l(unihiro 1972, Blake 19112 Kunihiro 1972

Nihei 1978 Nihei 1978, Blake 1982, Barsom 1985 ;7r-- K" - 111111110. AN Albrecht 1982 a EiEy

Albrecht 1980,1982a, 1983o,Yamada 1983b

Barsom 1983 Yamado 1983 b

Figure 78. Types offatigue tensile specimens. 101

oiled Beams

1640

200 2-YEAR L. 0.

CATEGORY A 20 Lu )- YEAR 2-YEAR cr

cn 10cr 0-YEAR cn a-YEAR Wa' RUNOUT CAT. A MEAN 15 Figure 79. Fatigue strength of 2-year weathered base metal specimens fabricated from Japanese SMA 50 and 2 5 2 SMA 58 weathering steels. [Kunihiro io - 1o6 1972] CYCLES TO FAILURE

ROLLED BEAMS

400 o 13 6O L.

200 C a- CATEGORY A 20 w Si L_._0. YEAR Cs 4-YEAR 4 IOO-

U) 0 U) (I) Si cr ir 0 0- YEAR I- U) a 4- YE AR " 50 Wa' RUNOUT CAT.A MEAN Figure 80. Fatigue strength of 4-year weathered base metal specimens fab- ______I I I ricated from Japanese SMA 50 and io 2 5 106 2 5 io' SMA 58 weathering steels. [Kunihiro CYCLES TO FAILURE 19721

Category A mean, as indicated by the fatigue notch factors, K1 18.3 percent (15 out of 82) of the corresponding data points 0.94 and 0.92, less than unity (Table 43). falling below the Category A allowable line. (See points 3 and The long life of the groove-welded specimens likely resulted in Figure 94.) from the extra care in welding caused by the element of com- The loss in life was about the same after 2 years and 4 petition between the seven fabricators. years of weathering. Two years and 4 years of weathering decreased the mean fatigue life of the base metal specimens by 57 percent and 54 percent, respectively, with 16.7 percent (11 out of 66) and 10.8 Manually Welded Stlffeners and Attachments percent (9 out of 83) of the corresponding data points falling below the Category A allowable line. (See Table 43 and points Albrecht, et al., reported the results of a study [Albrecht 1 and 2 in Figure 94.) 1982a, Friedland 1982] in which the following variables were The losses in life of the groove-welded specimens were 60 examined: (1) specimens with transverse stiffeners and 4 in. (102 percent and 63 percent, with 22.9 percent (16 out of 70) and mm) attachments; and (2) 0-year, 2-year, 2-year alternate, 4-

102

year and 4-year alternate weathering. The first specimen con- the service life would be 0.3 percent if he would accept the sisted of a 1-in. x %-in. x 13-in. (25-mm X 10-mm x 330- continuously weathered stiffeners. mm) long main plate to which two 1-in. X 4-in. X 2-in. (25- Fatigue cracking occurred in nearly equal numbers from mm X 7-mm X 50-mm) long transverse stiffeners were attached weld toes facing the sky or the ground. The effect of side of with 4-in. (7-mm) manual fillet welds. The other series consisted exposure on fatigue life was, in general, statistically insignificant. of a 2/2-in. x %-in. x 13-in. (64-mm x 10-mm x 330-mm) long main plate to which two V/4-in. X -in. X 4-in. (45-mm The fatigue test data for the specimens with 4-in. (102-mm) X 7-mm X 102-mm) long flat attachments were manually attachments suggest the following conclusions: welded with 3/ 6-in. (5-mm) fillet welds. The specimens were made of A588 Grade A steel. The so-called alternate weathering con- The mean fatigue life of the 0-year weathered 4-in. (102- sisted of repeatedly exposing the specimens for 6 months and mm) attachments (K1 = 1.95) was equal to that of Category C applying either one-fourth or one-eight of the estimated cycles transverse stiffeners (K1 = 1.95). It was longer than the life of to failure of the 2-year and 4-year weathered specimens, re- the 4-in. (102-mm) attachments welded to beam flanges that spectively. have become the basis for Category D [Fisher 1974]. The reason As an example, the fatigue test data for the transverse stif- for this behavior was the reduced degree of constraint provided feners are plotted in Figure 81. The following conclusions can by the main plate of the tensile specimens, as compared to that be drawn from the stiffener data: provided by the more rigid beam specimens. Still, the data gave additional information on the behavior of details with Category The mean fatigue life of the 0-year weathered stiffeners C fatigue strength. (K1 = 2.04) was slightly shorter than that for Category C beams Two-year and 4-year weathering decreased the fatigue life and girders with stiffeners welded to the web and flanges (K1 by 6 percent and 10 percent, respectively (points 9 and 10 in = 1.95) [Fisher 1974]. Figure 94). Two-year and 4-year weathering decreased the fatigue life Two-year alternate weathering increased the life by 8per- by 21 percent and 22 percent respectively. (See Table 43 and cent, whereas 4-year alternate weathering decreased it by 9 points 5 and 6 in Figure 94.) percent (points 11 and 12 in Figure 94). Two-year alternate weathering increased the life by 6 per- The loss in life was found by analysis of variance to be cent, whereas 4-year alternate weathering decreased it by 16 statistically insignificant at the 5-percent level. percent (points 7 and 8 in Figure 94). The effect of side of exposure on fatigue life was statistically All data points and, for the most part, the lower confidence insignificant. limits at two standard deviations from the mean fell above the Category C allowable line. The significance of the loss in fatigue life was examined Automatically Welded Stifteners with analysis of variance and with acceptance sampling [Ang 1975]. Two analyses were performed. In the first case, assuming In a second study, Albrecht, et al., tested A588 Grade B steel a 5 percent risk to the bridge supplier and ignoring the risk to specimens with transverse stiffeners identical to those of the first the bridge owner, the loss in life was found to be statistically study, except that they were welded by the automatic sub- insignificant. In the second case, assuming a 5 percent risk to merged-arc process [Albrecht 1980 and 1983a]. The weathering the bridge supplier and considering the risk to the bridge owner, times were 0-year, 3-year, 3-year alternate and 8-year. The al- it was found that the owner's risk of a fatigue failure during ternately weathered specimens were exposed for 3 years, fol-

£ 0 4jVjV A 40

° 200L ---- to

20 C, I

U) I•• L OO- Ye 10 Iii I ll L £2-Year Category C 1 UI 'n A 2-Year Alt 501- • 4Year Figure 81. Fatigue strength of 2-year O 4-Veor Alt - Runout - and 4-year weathered manually welded transverse stiffeners fabricated from A588 steel. [Albrecht 1982a, Friedland 107 5lO 10 1982] CYCLES TO FAILURE

103

lowed by alternate one-quarter-life cycling and 6-month weld toes facing the sky or the ground are indeed well inter- weathering. spersed in Figures 83 and 84. Figures 82 and 84 depict the gradual degradation of fatigue The data were compared with the current specification re- strength with exposure time. The 0-year weathered specimens quirements in three ways. First, the data points were plotted (K1 = 1.67) had a fatigue strength closer to that of the Category against the Category C allowable line. Figure 85 shows that all B mean for welded beams (K = 1.49) than to Category C mean 62 points fell above this line. for manually welded stiffeners (K1 = 1.95), because the superior Secondly, the lower confidence limits at two standard devia- quality of the automatic submerged-arc welds increased the tions to the left of the mean were compared. For transverse crack initiation portion of the fatigue life. Figure 83 compares stiffeners manually welded to beams and girders, Table 39 gives the 3-year data against the "0-year rhomboid." The latter is the b = 12.681, m = 3.097, and s = 0.158. The corresponding space between the two-standard-deviation confidence limits of values for the automatically welded stiffeners are given in Table the 0-year data. As can be seen, the 3-year data fell in the left 4.0. Substituting these sets of values into Eq. 30 yields the four half of the 0-year rhomboid. Figure 84 shows the 8-year mean lower confidence limits drawn with dashed lines in Figure 86, falling on the lower confidence limit of the 0-year data, with along with the solidly drawn limit for stiffeners welded to beams one-half of the points falling short of the rhomboid. The fatigue and girders (Category C*), and the limit for 2-in. (51-mm) strength degraded at a decreasing rate with increasing time of attachments (Category Q. The lines in Figure 86 are labeled exposure. in descending order of life. The four limits for the data fell above The effect of side of exposure on fatigue life was statistically the Category C allowable line. The 8-year limit fell below the insignificant. The data points for the specimens that failed from limit for stiffeners welded to beams and girders.

,-0-Year Rhomboid 40

200 -ix_-- o - A— - - a. 0-Year 20

Ui I tOO ______(z 2s ] 4 I Co 10 Co Ui

Figure 82. Fatigue strength of 0-year weathered automatically welded transverse stiffeners fabricated from A588 105 steel. [Albrecht 1980 and 1983a] CYCLES TO FAILURE

0 a-

w 20 C, z 4

C') C,) Ui a. I— U)

Figure 83. Fatigue strength of 3-year weathered automatically welded trans- 2 5 2 5 verse stiffeners fabricated from A588 105 io6 107 steel. [Albrecht 1980 and 1983a] CYCLES TO FAILURE

104

k ci 60

- 40

U, 8-Year 20 z Ui Cr 100 - (0 U) - U) Ui

I— U) - Skyward

- ' Groundword Runout 5 Figure 84. Fatigue strength of 8-year I III I I weathered automatically welded trans- 2 5 2 5 10 5 to6 verse stffeners fabricated from A588 CYCLES TO FAILURE steel. [Albrecht 1980 and 1983a]

[]

tL •• CC t4• • 200 Neon for Stitfeners Welded to Beams and Girders a- o o. w Ui ' C, 00 z 4

U) U) Ui yeor Ir I— 0 3-Year Category C Cn T- 50 0 8-Year

5 Figure 85. Comparison of all data for automatically welded transverse stiff: o I I I JO eners fabricated from A588 steeL [Al- 10 10 brecht 1980 and 1983a] CYCLES TO FAILURE

LIMITS IN DESCENDING ORDER 400 0-YEAR 60 3-YEAR AND ALTERNATE STIFFENERS ON BEAMS AND GIRDERS 8-YEAR 40 51mm ATTACHMENTS - 200 CL

200

(D IOO-

C') U) Ui Category C I— Cl) 50

5 Figure 86. Comparison of lower con- I fidence limits for automatically welded L I I 2 transverse stiffeners fabricated from IO 106 A588 steel. [Albrecht 1980 and 1983a] CYCLES TO FAILURE

105

Thirdly, safety factors were compared. They were defined, STRESS RANGE(ksi)

for the purpose of this evaluation, as the ratio between the mean 0 10 20 30 40 line of a data set and the Category C allowable line. Figure 87 5, shows that weathering reduced the safety factor on life of the

automatically welded stiffeners from about 4.0 for the 0-year - 0-Year data to 2.5 for the 3-year and alternate data, and 1.9 for the 8- Li La year data. These drops correspond. to the reported 42 percent -I and 54 percent losses in life, respectively. The mean safety factor z 0 for the 8-year data was much lower than the safety factor of Alternate 0 Mean for Stiffeners Welded 2.5 for nonweathered stiffeners manually welded to beams and to Beams and Girders 0 girders. In summary: I - 3-Year I-. Three-year, 3-year alternate, and 8-year weathering re- (Li duced the life by 42, 42, and 54 percent, respectively. (See Table U- (I) 43 and points 13, 14, and 15 in Figure 94.) JI The data points and the lower confidence limits fell above - Category C Allowable Line the Category C allowable line for 2-in. (51-mm) attachments. This happened for two reasons: (1) the 0-year weathered specimens had a fatigue strength close to that of Category B; and (2) the standard deviation for the weathered specimens was 0 100 200 300 smaller than for the nonweathered specimens. STRESS RANGE (MPa) The mean safety factor on life dropped from about 4.0 for the 0-year data to 1.9 for the 8-year data; that is, below the Figure 8Z Comparison ofsafetyfactorsfor automatically welded safety factor of 2.5 for the manually welded transverse stiffeners transverse stiffeners fabricated from A588 steel. [Albrecht 1980 that are part of the data base for Category C. and 1983a] The lower confidence limit continued to drop with increasing weathering time, although at a smaller rate. This suggests that the degradation of fatigue life slows down with time, but does not stop.

The fatigue test data for the groove-welded plates with weld reinforcement left in place are plotted in Figure 88. The data Base Metal and Groove Welds indicate the following:

Barsom quoted Blake as having tested the following three The nonweathered specimens had a mean fatigue strength types of A588 Grade A steel plate specimens that were weath- (K1 = 1.39) higher than, the mean of the Category B welded ered up to 6 years: (1) base metal, as received from the mill; beams (K1 = 1.49). (2) blast-cleaned base metal; and (3) groove welds [Blake 1982, Two-year weathering reduced the mean strength Barsom 1983]. The base metal specimens were subjected to stress (K1 = 2.07) by 72 percent to less than the mean of the Category ranges varying from 45 to 60 ksi (310 to 415 MPa). The results C transverse stiffeners (K1 = 1.95). The data are represented suggest the following conclusions: by point 20 in Figure 94. For those specimens cycled at 45-ksi (3 10-MPa) nominal The regression analysis of the data, excluding the runout stress range, local yielding at the toe of the weld reinforcement specimens, indicated that at 500,000 cycles the mean strength greatly reduced the fatigue strength before weathering and, of the nonweathered as-received steel (K1 = 1.00) was identical hence, also the loss in life after weathering. to the mean for Category A rolled beams (Kf = 1.00). The nonweathered blast-cleaned steel had higher strength (K1 = 0.84) because the residual surface compression stresses that were Transverse Stiffeners and LongitudInal Gussets introduced by blast cleaning increased the crack initiation life. Three-year and 6-year weathering reduced the life of the Yamada and Kikuchi examined the effect of 0-year, 2-year, as-received steel specimens by 28 percent. Two-year and 4.5- and 4-year weathering on the fatigue behavior of two types of year weathering reduced the life of the blast-cleaned specimens specimens fabricated from Japanese SMA 50 weathering steel by 24 percent and 47 percent respectively. (See points 16-19 [Yamada 1983b]. One type consisted of a 3.15-in. X 0.39-in. in Figure 94.) X 24-in. (80-mm x 10-mm x 600-mm) long plate to which The loss in life was much larger for the specimens cycled two 3.15-in. X 0.39-in. X 2.36-in. (80 mm X 10 mm X 60 at or below the yield point than for those cycled above the yield mm) long transverse stiffeners were welded with 0.24-in. (6- point. If the data points for specimens cycled at more than 50- mm) manual fillet welds. The other type of specimen consisted ksi (345-MPa) stress range were excluded, as they should be, of a 3.15-in. x 0.39-in. x 28-in. (80-mm X 10-mm x 700- the losses in life would be 69, 68, and 76 percent for the 3-year mm) long plate to which two 2-in. x 0.39-in. x 4-in. (50-mm and 6-year as-received steel, and the 4.5-year blast-cleaned steels, X 10-mm x 100-mm) longitudinal gussets were welded all respectively. (See points 16a, 17a, and 19a in Figure 94.) around with 0.24-in. (6-mm) manual fillet welds. 106

Figure 88. Fatigue strength of groOve- welded plates fabricated from A588 CYCLES TO FAILURE steel. [Blake 19821

The fatigue test data for the transverse stiffeners are plotted Therefore, the scatterband of the weathered specimens data in Figure 89. Two-year and 4-year weathering increased the expanded upward. But the lower confidence limit for the weath- fatigue life of the stiffeners by 51 percent and 91 percent, rè- ered specimen data remained unchanged from that for the non- spectively. (See points 21 and 22 in Figure 94.) The gain in weathered specimens. The measured pit depths on the surface fatigue strength after weathering pushed the data points for of the base metal increased from 0.004 in. (0.11 mm) before stiffeners above the Category B allowable line. weathering to 0.008 in. (0.21 mm) after 4-year weathering. Pit The data for the longitudinal gussets showed 70 percent and depths were not measured near the weld toes where the fatigue 39 percent increases in fatigue life after 2-year and 4-year weath- cracks initiated. ering, respectively. (See points 23 and 24 in Figure 94.) An additional 10 specimens with transverse stiffeners were Yamada attributed the gain in life to weld-toe rounding due cut from a temporary SMA 50 steel bridge that had been used to corrosion which reduced the stress concentration factor and in Tokyo for 5.5 years. The specimen dimensions were identical thus increased the crack initiation life. Measurements showed to those of the transverse stiffeners fabricated for the research that the root radius at points along the toe of the stiffener weld program. The fatigue test data fell above the Category C allow- increased from 0.028 in. (0.7 mm) for nonweathered to 0.060- able line. Because there are no 0-year control data in this case, 0.120 in. (1.5-3.0 mm) for weathered specimens. According to it cannot be determined if the bridge specimens gained or lost Yamada, a gain in life could not always be expected because life after weathering. the weld toe radii did not increase along the full weld length.

400 60

2-Year 140

- 0 0. 0. -20

Ui I00 4 C, 0 0-Year z U) A 2- Year 10 < 4-Year Category C cc (n 50 U) J1 LU 4! -~ 5 U)I—

Figure 89. Fatigue strength of transverse stiffenersfabricatedfrom SMA 50 105 2 5 1 06 2 -. 5 steel. [Yamada 19831 CYCLES TO FAI LURE 107

Bead Welds and Groove Welds Fisher 1970] overlap, the groove welds in the three environments had 2.0, 3.2, and 3.2 times longer mean fatigue lives than the Barsom [1983] reported data for 86 groove-welded and over Category A rolled beams. Even at the 70-ksi (483-MPa) stress 47 welded specimens that were fatigue tested after 11-years range, that is 12 ksi (83 MPa) higher than the measured 58-ksi exposure in rural, urban-industrial, and moderate marine en- (400-MPa) yield strength of the A242 and A588 steel specimens vironments. The reinforcements of the bead and groove welds tested in Barsom [1983], twice as many specimens fell above were left in place. The specimens were made of A242 Type 1 than below the Category A allowable line. and A588 Grade A steels. The lack of nonweathered specimen tests and the unlikely Figure 90 shows, as an example, the fatigue test data for the results make the data unsuitable for determining the effect of groove-welded specimens made of A588 steel. Nine specimens weathering on fatigue. tested at 70-ksi (483-MPa) stress range had lives shorter than 100,000 cycles. For this reason, their data points could not be plotted in the figure, but they were included in the regression Cover Plates analysis. Also shown, for comparison, is the Category A rhomboid. It encloses the 95.7 percent confidence space rolled beam Fisher weathered for 2 years one coverplated W36 beam fab- fatigue test data from which the Category A allowable line was ricated from A588 steel [Fisher 1983a]. The coverplated tension derived [Fisher 1970]. The findings are as follows: flange faced the sky during weathering and the ground during fatigue testing. To artificially accelerate the weathering process, Because no 0-year weathered specimens were tested, the the cover plate ends were packed with rock salt (NaC1) once a effect of weathering on fatigue life cannot be determined from month, a procedure that caused the ends to corrode very se- the data. verely. Fatigue cracks initiated at the root of the end welds and The groove welds had about equal fatigue lives after 11 propagated across the full cover plate width at both ends in years of weathering in the three environments, as indicated by 8,300,000 and 8,800,000 cycles of 6-ksi (41-MPa) stress range. the nearly identical fatigue notch factors (Kf = 0.95 - 0.96) The test was then discontinued. Root cracks are usually a sign in Table 43. of high fatigue strength. In this case, however, the two weathered The data points fell above the Category A allowable line, cover plate ends had, on average, shorter fatigue lives than those although groove welds and bead welds are Category C details. on nonweathered beams that failed from weld toe cracks after 10,000,000 cycles or more. The conditions that existed in these The reference [Barsom 1983] data raise questions about the tests are not relevant to bridges for the following reasons: (1) validity of the results. Bead and grooye welds with reinforcement salt in the quantities deposited on the top flange of the upside- not removed have Category C strength according to the down beam cannot adhere to the underside of the bottom flange AASHTO specifications. Yet, the reference [Barsom 1983] welds of a bridge girder; (2) artifical weathering cannot simulate nat- had Category A strength, that is 7.3 times longer allowable ural weathering; and (3) bridges would not be built from non- fatigue lives than Category C. At the 40-ksi (276-MPa) stress painted weathering steel in an environment like the one created range level in Figure 90, where the data [Barsom 1983 and for the test.

- 100 600L,—Marine I I rUrbon Industrial

400 O zo 00 60 0 to 0

0 QA, 40

200 'I Ui I 20 z I Ui z U) A Rural u, L cr Ui I 0 Urban-Industrial I- It) u I— I u h- 0 Marine Ui I— U,

Figure 90. Fatigue strength of 11-year 10 06 groove-welded specimens fabricated from A588 steel. [Barsom 19831 CYCLES TO FAILURE 108

NONATMOSPHERIC CORROSION RESISTANT specimens were analyzed in this study in terms of the tension STEELS portion of the stress cycle (f/2). The stress reversal data for the groove-welded specimens were analyzed in terms of the full Three Japanese studies examined the effect of weathering on stress range because residual tensile stresses elevate the applied the fatigue life of specimens fabricated from rolled steel for stresses. All of weathered specimens were fatigue tested to failure welded structures supplied to Japanese Industrial Standard after 3-years exposure. As an example, the data for the notched G3 106, designation SM 50 and SM 58 [JIS 1977]. These steels plate specimens, tested at R = 0.0, were plotted in Figure 91. do not have the atmospheric corrosion resistance of their SMA The findings are as follows: counterparts [JIS 1981]. They are comparable to ASTM A441 steel. The results of all regression analyses are given in Table The fatigue strength of the nonweathered notched speci- 41. The calculations of change in stress range and life due to mens (K1 = 0.69 for R = 0.0) exceeded that of Category A. weathering are given in Table 43. Three-year weathering reduced the fatigue life by 74 percent The following types of specimens, shown in Figure 78, were and 58 percent for R = 0 and R -1, respectively. (See tested: base metal [Kunihiro 1972], groove weld ground flush points 29 and 30 in Figure 94.) [Kunihiro 1972], groove weld with reinforcement not removed Three-year weathering reduced the fatigue life of the [Nihei 1978], notched plate [Nihei 1978], transverse stiffeners groove-welded specimens by 44 percent and 67 percent for R [Yamada 1983b], and longitudinal gussets [Yamada 1983b]. The = 0 and R = - I, respectively (See points 31 and 32 in Figure data are examined in the following by investigator. 94.)

Base Metal and Groove Welds Transverse Stlffeners and Longitudinal Gussets

Kunihiro, et al., repeated for SM steels the same experimental Yamada and Kikuchi repeated for SM steels the same ex- program described earlier under "Weathering Steels" for SMA perimental program described earlier under "Weathering Steels" steels [Kunihiro 1972]. The plots of the data for the SM steels for the SMA steels [Yamada 1983b]. The plots of the data were were very similar to those for SMA steels shown in Figures 79 similar to that shown in Figure 89. The findings are as follows: and 80. The findings are as follows: Two-year and 4-year weathering increased the fatigue life The mean fatigue lives of the nonweathered base metal of the stiffeners by 120 percent and 91 percent, respectively. and groove-welded specimens were slightly longer than the (See points 33 and 34 in Figure 94.) Category A mean, as indicated by the fatigue notch factors, Four-year weathering increases the fatigue life of the gus- Kf = 0.98 and 0.97, less than unity (Table 44). Only 4.8 percent sets by 11 percent. (See point 35 in Figure 94.) No 2-year (3 out of 62) and 4.0 percent (2 out of 50) of the data points, weathered gussets were tested. respectively, fell below the Category A allowable line. Both types of specimens exhibited about the same fatigue The reasons for the observed gains in life, given by the in- strength. vestigators, were already cited under "Weathering Steels." Two-year and 4-year weathering reduced the fatigue life of the base metal specimens by 58 percent and 62 percent, respectively, with 24.6 percent (17 out of 69) and 18.8 percent EFFECT OF PITTING (16 out of 85) of the corresponding data points falling below , the Category A allowable line. (See Table 44 and points 25 and The observed losses in life of specimens fabricated from weath- 26 in Figure 94.) ering steels and nonatmospheric corrosion resistant steels are The losses in life were 60 percent and 64 percent for the likely caused by pitting, as the work by Hiam and Pietrowski groove-welded specimens, with 23.9 percent (17 out of 71) and shows [Hiam 1978]. In that study, specimens were fabricated 28.4 percent (25 out of 88) of the corresponding data points from a 0.06 percent C hot-rolled low-carbon (HRLC) mild steel falling below the Category A allowable line. (See points 27 and of 36 ksi (250-MPa) yield strength and a silicon semikilled 28 in Figure 94.) columbium-strengthened HSLA steel of 50-ksi (345-MPa) min- The loss in life was about the same after 2-year and 4-year imum specified yield strength, designation Dofascoloy 50W. The weathering. specimens were fatigue tested in the following conditions: (1) hot rolled surfaces; (2) laboratory pitted; (3) corroded under vehicle; and (4) corroded surfaces removed. The following con- Notched Plates and Groove Welds clusions were reported:

Nihei, et al., tested flat plate tensile specimens fabricated from Fatigue cracks initiated from rust pits. SM 50 A steel [Nihei 1978]. The principal variables in the Pit shape and type of fatigue failure were similar for spec- experiment were: (I) notched specimens versus groove-welded imens that were laboratory pitted or corroded under a vehicle. specimens; (2) minimum-to-maximum stress ratios of -1.0 and Both produced similar losses in fatigue life. 0.0; and (3) 0-year versus 3-year weathering. The notches were The loss in life increased as the pit depth increased. 0.5 in. (12.5 mm) deep and had a 1.37-in. (35-mm) radius, for Removal of the corroded surfaces restored the original a theoretical stress concentration factor of 1.22. The weld re- fatigue strength of the specimens. inforcement on the full penetration groove welds was not re- The loss in life was greatest at low-strain amplitude, that moved. The stress reversal data, R = - 1, for the notched is in the long-life regime (100,000 to 10,000,000 cycles). 109

00 600 a

50 0- Year E 3-Year 200 Cat.00rv A

Ui 100 CC 0 0-Year C', 3-Year ', 60 Ui - Runout I— 40 U,

2 5 2 105 2 5 10€ 10 7 CYCLES TO FAILURE

Figure 91. Fatigue strength of 3-year weathered notched plate specimens fabricated from SM 50 steel. [Nihei 19781

6. There was no significant difference between the losses in PIT DEPTH (mils) life for the two steels. 79 . . It Figures 92 and 93 give the results of the pit depth studies on the HRLC and HSLA steels, respectively. Both show the same trend of reduced fatigue life as pit depth increased. The summary of loss in life, given in Table 42, reveals that the two steels behaved alike, with losses of 78 percent and 84 percent at 0.009- in. (230-sm) pit depth [Hiam 1978]. In comparison, low-alloy steels reached that pit depth after less than 3-years exposure in 11 industrial Bayonne, N.J., as shown in Figure 70 [Copson 1960]. The following data from other sources indicate the degree of pitting penetration that can be expected under various conditions: (1) 15 mils (370 pPm) for low-alloy steel ideally exposed 18.1 years in industrial Bayonne, N.J. [Copson 1960]; (2) 28 miles (707 m) for low-alloy steel ideally exposed 17.1 years in marine Block Island, R.I. [Copson 1960]; and (3) up to 250 mils (6,000 m) for weathering steel bridges in Michigan ex- [Ii posed for 15 years to severe deicing salt environments [Culp 1980]. Evidently, pitting penetration of environmentally exposed weathering steel can be much larger than that of the reference [Hiam 1978] panels. Since pitting penetration increases with time (Fig. 70), and loss in life increases with pit depth (Figs. 92 and 93), one would 4 [II expect a continuing degradation of the fatigue strength of non- 0 100 200 300 400 painted weathering steel bridges, although at a declining rate. The data [Albrecht 1983a] for automatically welded stiffeners PIT DEPTH (pm) show, in fact, a continuing degradation in mean fatigue strength (Figs. 82 to 84) and lower confidence limit (Fig. 86) in 3-year Figure 92. Fatigue life versus pit dej ih for HRLC mild steel. and 8-year weathered specimens. Sixteen-year data from current studies at the University of Maryland will not be available until 1992. factor (Eq. 33) of the corresponding nonweathered control specimens. The data points for the 24 A588 and SMA weathering COMPARISON OF WEATHERING FATIGUE DATA steel series are shown with triangular symbols, those for the 11 SM steel series with circular symbols. Each point represents the The losses in life for the 24 series of weathering steel specimens average loss in life for one series of tests. These 35 points are and the 11 series of nonatmospheric corrosion resistant steel based on an analysis of 1,431 fatigue tests, not counting over specimens are summarized in Tables 43 and 44, respectively. 132 fatigue tests performed by Barsom for which there were no The losses are plotted in Figure 94 against the. fatigue notch 0-year weathered data. The data base for determining the effect

PIT DEPTH (mils)

E 15 E' 2 4' 40 0t o2 C) t co 0'0! c 4, o 0— 00 00 0 0 c- O) 0— 00 IOC

AA588B SMA - NI9 OSM 80- \ AH 1 - 620 -18u 17 82 32—.' 60 - 4Tc'NAlS Weathering and if2703 Corrosion Fatigue - 2A125\ "N 19 Eq 40- 031 ,l3,l4 4(, \ I \I 16,17 A 20 j,weatherin9 - 10NI2- ' — -Eq. 0 ------_. ------II 667 035 20.

40 A24 0 100 200 300 400 PIT DEPTH(pm) 216 -60. Figure 93. Fatigue live versus pit depth for HSLA steel, .Dofas- - A23 coloy 50W - 80 0 - 34 Q22 oc 33 • I A A

Table 42. Loss in fatigue life due to corrosion pitting. [Warn 1978] FATIGUE NOTCH FACTOR, K

HRLC Steel HSLA Steel Figure 94. Loss in life due to weathering fatigue. Pit Depth Fatigue Loss in Fatigue Loss in m) Life Life Life Life (cycles) (5) (cycles) (5)

0 740,000 -- 292,000 -- The comparison of the results from the 35 series of tests, 50 529,000 29 194,000 33 shown in Figure 94, leads to the following conclusions: 100 378,000 49 129,000 56 230 164,000 78 46,900 84 The loss in life is highest for base metal and continuously decreases as the stress concentration of the detail increases. In other words, the higher the stress concentration- of a detail, the less rust pitting reduces the life. of weathering on fatigue is very large. There are currently about Atmospheric exposure affects weathering steels and regular three times as many tests as the 499 data points that were used steels alike. to derive the Category A to E allowable lines [Fisher 1970 Fisher 1974]. Yamada's results [1983b] are not consistent with those ob- The vertical grid lines in Figure 94 were drawn at the values tained by the Japanese Ministry of Construction, University of of K1 for the Category A to E mean lines, given in Table 39. Maryland, and U.S. Steel [Kunihiro 1972; Albrecht 1980, 1982a, This allows one to compare the data for the weathered specimens and 1983a; Blake 1982]. His results show large gains in life, with the data base for the current fatigue specifications. Note whereas all others show losses in life. The weld-toe rounding, that the data for the 35 series were plotted against the actual to which Yamada attributed the gains in life, did not seemingly experimentally determined fatigue strength of the nonweathered occur in the specimens tested by others. The scatter in his series specimens, not against projected strength implied by the of tests (points 21-24 and 33-35 in Figure 94) is much larger AASHTO classifications. For example, the points for the higher than other investigators have found (points 1-4, 5-8, 9-12, strength automatically welded stiffeners were plottedat K1 = 13-15, 16-19, and 25-28). The weld-toe rounding may have 1.67, those for the lower strength manually welded stiffeners at been caused by galvanic corrosion between the weathering steel Kf = 2.04. base and a less noble weld metal. - 111

Table 43. Effect of weathering on fatigue life of specimens fabricated from weathering steels.

Synbol in Stress Range Loss in Loss in Fatigue Figure Type of Weathering at 500,000 Stress Fatigue Notch 94 Reference Detail Time Cycles Range Life Factor rw NW Kf (Dears) (MPa (% (U)

Kunihiro Base metal 0 373 -- -- 0.94 1 1972 2 287 23 57 1.22 2 4 294 21 54 1.19 Kunihiro Groove welds 0 380 -. -- 0.92 3 1972 ground flush 2 285 25 60 1.23 4 4 278 27 63 1.26

Albrecht Manually welded 0 172 -- -- 2.04 5 1982a transverse 2 160 7 21 2.18 6 stiffeners 4 159 8 22 2.19 _2a _6a 7 2 alt. 175 2.00 8 4 alt. 163 5 16 2.15 Albrecht 102-on 0 179 -- -- 1.95 9 1982a attachnents 2 177 -- 6 1.98 10 4 173 4 10 2.03 _8a -- 11 2 alt. -- -- 12 4 alt. -- -- 9 -- Albrecht Automatically 0 210 -- -- 1.67 13 1980, welded transverse 3 177 16 42 1.88 14 1983a stiffeners 3 alt. 177 15 42 1.98 15 8 165 21 54 2.13

Blake Base metal 0 352 -- -- 1.00 16 1982 as received 3 318 10 28 1.10 17 6 318 10 28 1.10

Blake Base metal 0 418 -- -- 0.84 18 1982 blast cleaned 2 382 9 24 0.92 19 4.5 343 18 47 . 1.02 Blake Groove welds 0 252 -- -- 1.39 20 1982 2 170 33 72 2.07

Vainada Transverse 0 193 1.82 --14 a _51a-- 21 1983b stiffeners 2 220 1.60 22 4 -- -- 9 1 a --

Yaniada 100-rn gussets 0 150 -- -- 2.83 23 1983b 2 177 _18a 70a 1.98 _11a 39a 24 4 167 2.10

Borsom Groove welds 11_Rb 364 -- -- 0.96 1983 11-UI 365 -- -- 0.96 11-M 369 -- -- 0.95 Barsain Groove welds 11-9 366 -- -- 0.96 1983 11-UI 372 -- -- 0.94 11-M 414 -- -- 0.85 Barson Bead welds 11-9 369 -- -- 0.95 1983 11-UI 398 -- -- 0.88 11-9 419 -- -- 0.84 Barsnnc Bead welds 11-9 1983 11-UI 11-M

Notes

Gain in life and stress range.

B = rural, 81= urban industrial, M = moderate marine. Fatigue test data could not be obtained from the investigator.

On the basis of the data available in 1981, Albrecht had drawn been ignored because of the uncertainty in using the stress am- the.following upper-bound envelope for relative loss due to plitude, f, = 0.5 J, to calculate K1 for the notched plate spec- weathering [Albrecht 1982a, Albrecht 1983a]: imens tested under stress reversal, R = -1 [Nihei 1978]. Fatigue test data reported since 1981 for base metal (points l6a, = 1 - 0.38 Kf (34) 17a, and 19a), groove welds (point 20) and stiffeners (point 15) have yielded five more points that exceed the losses predicted by Eq. 34. In light of the new data, this equation now appears Equation 34 is based on data in the region of 0.69 . exhibit after. 50 years of service. Present indicatiois are that the long-term losses will be greater. w 0.0 (35) LN = The weathering fatigue data are used in Chapter Eight, in conjunction with the corrosion fatigue data, to determine the Equations 34 and 35 are plotted with dashed lines in Figure allowable stress ranges for fatigue design of weathering steel 94. The only data point (30) above Eq. 34 in Figure 94 had structures. 112

Table 44. Effect of weathering on fatigue life of specimens fabricated from SM steels with no atmospheric corrosion resistance.

Stress Range Lossin Loss in Fatigue Synbol in Type of Weathering at 500,000 Stress Fatigue Notch Figure Reference Detail Time Cycles Range kife Factor N K 94 °rw W (Years) (MPa) (Xl (6)

Kunihiro Base metal 0 359 -- -- 0.98 25 1972 2 274 24 58 1.28 26 4 265 26 62 1.32

Kunit.ro Groove weld 0 362 -- -- 0.97 27 1972 ground flush 2 272 25 60 1.29 28 4 264 27 64 1.33

Nihei Notched plate 0 506 -- -- 0.69 29 1978 3 332 34 74 1.05

Nihei Notched plate 0 257 -- -- 1.36 179b 30 1978 0 = -1 3 196 24 58

Nihei Groove weld 0 302 -- -- 1.16 31 19)8 as welded 3 252 17 44 1.39 396z 088c Nihei Groove weld as 0 -- -- 125c 32 1978 welded R = -1 3 279 30 67

Yamoda Transverse 0 194 -- -- 1.81 _120a -- 33 1983b stiffener 2 -- -- _91a -- 34 4 -- --

Yanada 100-rn gusset 0 160 2.19 --3 a --11a 35 1983b 4 165 2.12

Notes:

Gain in fatigue life Dosed on stress amplitude Bused on stress range

CHAPTER EIGHT

FATIGUE DESIGN

The total fatigue life consists of a crack initiation phase plus The first two types are obtained from tests of standardized a crack propagation phase. Weathering, as discussed in Chapter specimens. They provide information on crack initiation and Seven, reduced only the crack initiation phase, because the spec- crack propagation in a given structural steel. The combined data imens were fatigue tested after the end of the weathering period. can be used, in conjunction with mathematical models of ini- In reality, weathering steel bridges are exposed to the environ- tiation and propagation lives to determine the effect of corrosion ment during both phases of fatigue cracking. The latter effect, on the total fatigue life of a detail. termed corrosion fatigue, affects both the crack initiation and The third type of data is obtained by fatigue testing a detail propagation phases. to failure in the corrosive environment of interest. Such tests Chapter Eight first presents the available corrosion fatigue give directly the effect of corrosion on the total fatigue life. The data. This is followed by an estimation of total loss in life and three types of data are examined in the following to assess the an examination of allowable stress ranges. corrosion fatigue life of weathering steel bridges.

CORROSIVE FATIGUE Corrosion Fatigue Crack Propagation

The effect of a corrosive environment on fatigue life can be Barsom and Novak published CFCP data in the form of plots assessed by comparing data collected in clean air and in the of crack growth rate, da /dN, versus range of stress intensity environment of interest. The following types of data are avail- factor, AK, for A36, A588 Grades A and B, and A5 14 steels. able: They were tested in air, distilled water, and 3 percent sodium chloride solution [Barsom 1977]. They presented no analysis of Corrosion fatigue crack initiation (CFCI) life. the data and have since discarded the raw data. Therefore, a Corrosion fatigue crack propagation (CFCP) rate. total of 1,651 data points had to be digitized from the plots of Corrosion fatigue S-N (CFS-N) data. the reference [Barsom 1977] for analysis in the present study

113

[Yazdani 1983]. The log-log mean regression lines were calcu- da - 3.77 iO' (A)332° (37) lated for all steels and environments using the Paris equation: - x

was significantly higher than the crack growth rate for A588 da steel in the air environment = C MC" (36) dN do - — 1.65 x 10_12 (..K)3475 (38) TN in which C and n are material constants. The statistical difference between any two mean lines was determined with T-tests Substituting AK in units of MPa-J gives da/dN in units of at the 5 percent level of significance. rn/cycle. The aqueous environments increased the crack growth The crack growth rate data for A588 steel in air and aqueous rate by a factor of 1.48 at the lowest value, i.K = 14.8 environments are plotted in Figures 95 and 96. The results of MPa,J (13.5 ksiji.), at which data were collected. the regression analyses and statistical comparisons, given in 3. The statistical conclusions for A588 steel, drawn from Table 45, lead to the following conclusions: items 1 and 2, are also valid for A36 and AS 14 steels (Table 45). The difference in crack growth rate of A588 steel in dis- 4. The difference in crack growth rate for A36 and A588 tilled water versus 3 percent sodium chloride solution was in- steels in air or aqueous environments was insignificant (Table significant. The data for the two aqueous environments were, 45). therefore, combined for comparison with the air data. The crack growth rate for A588 steel in aqueous environ- Including the previously rnentioned data from Barsom (1977), ments Yazdani and Albrecht collected a total of 3,254 data points for

&c(MPa T) AK(MPa vTh)

0 2 5 10 2 5 100 2 11 IC) 9 IC)O

T 0 0

U. '

2: z —c 0 .- -. C 0 0 U LJ (0 L 2: Cr I I-- I— 0 C 2: 2: N 2 () . 1< 2: 2: U U

CD 0 0

C 0

, K(ksi v') K ( k s i V1. )

Figure 95. Crack growth rates for A588 steel in air. [Barsom Figure 96. Crack growth rates for A588 steel in aqueous en vi- 1977] ronments. [Barsom 19771 114

Table 45. Regression analysis and statistical comparison of crack growth rate data for A36, A588 Grades A and B, and A514 steels.

Steel and No. of Results of Standard' I Id Variation Environmenta Data Regression Analysisb Deviation Statistic Critical Significant Points c n

A36 H20 vs. 29 5.58 E-13 3.877 .0557 2.06 +2.08 NO A36 NaCl 161 4.33 E-12 3.152 .1370 1.59

A588 H20 vs. 44 1.10 E-12 3.610 .0742 1.95 +2.00 NO A588 NaCl 271 4.18 E-12 3.300 .1593 0.54

A514 H20 vs. 58 3.86 E-11 2.482 .0851 0.45 +1.97 NO A514 NaCl 426 7.11 E-11 2.407 .1813 0.73

A36 Air vs. 92 1.05 E-12 3.609 .0811 9.25 +1.99 YES A36 Aqueous 190 4.71. E-12 3.221 .1336 8.46 Aqueous faster

A588 Air vs. 168 1.65 E-12 3.475 .0732 2.73 +1.98 YES A588 Aqueous 315 3.77 E-12 3.320 .1552 1.83 Aqueous faster

A514 Air vs. 372 5.28 E-12 3.026 .0883 27.15 +1.96 YES A514 Aqueous 484 6.02 E-11 2.420 .1563 11.27 Aqueous faster

A36 Air vs. 92 1.05 E-12 3.609 .0811 1.15 +1.99 NO A588 Air 168 1.65 E-12 3.475 .0732 0.38

A36 Aqueous vs. 190 4.71 E-12 3.221 .1336 0.77 +1.97 NO A588 Aqueous _315 3.77 E-12 3.320 .1552 1.31

Notes: H2O = distilled water; NaCl = 3% sodium chloride solution. Substituting K values in units of MPaiWyields crack growth rates, da/dN, in rn/cycle. Standard deviation of log (da/dN). Critical value at 5% level of significance.

crack growth rate of mild steel (A36), high strength low alloy . da .344 = 54 < 10-12 (.K)3 steels (A588, X-52), and quenched and tempered steel (A514) dN in air and aqueous environments [Yazdani 1983]. The aqueous environments consisted of distilled water and a solution of 50 For mild and HSLA steels in aqueous environments: dium chloride in distilled water. After a statistical analysis of the effects of type of steel, loading, environment and source, the da 12 data were combined into four sets. Correspondingly, the follow- = 4.16 X 10- (.K)3•279 dN ing crack growth equations were obtained for mild and HSLA steels in air: For quenched and tempered steels in air: 115

do = A K (M Pa ( Vii) 2.27 X 10_I1 ()2.534 (41) 0 0 • bC For quenched and tempered steels in aqueous environments:

da = 6.00 x 10- (X)2420 (42) A Mild 8 HSLA Air ' The pairs of air and aqueous lines for each group of steels, £ Mild B HSLA Aqueous - plotted in Figure 97, are practically parallel. As in the case of 0 QT Air the Barsom [1977] data for A588 steel alone (Eqs. 37 and 38), QT Aqueous the crack growth rates for the combined mild and HSLA steels and for the quenched and tempered steels are higher in the aqueous environments than the corresponding growth rates in air. Both sets of equations suggest that structural details on nonpainted steel bridges have shorter crack propagation lives 0 than those on painted bridges. Fisher, on the other hand, stated that the investigation by Barsom and Novak [Barsom 1977] had concluded that the CFCP life of steel bridge components under actual operating I- (wet and dry) conditions in the aqueous environments tested 0 should be equal to or greater than their fatigue life in room temperature air environments [ Fisher 1983b]. This conclusion was based on an interrupted test of a single AS 14 steel specimen in salt water. In this test, the subsequent decrease in crack growth rate had originally been attributed to an overnight air 0 exposure of the specimen during an equipment malfunction. However, according to Novak, the crack growth delay is now believed to have been caused by an overload accidentally applied a' on the specimen when the equipment malfunctioned [Novak '0 1983a, Albrecht 1983c]. Single overloads are known to delay 2 34567891C crack growth under subsequent cycling at lower stress ranges A K ( ksi Vi) [Abtahi 1976]. The greatest part of the fatigue life 'of bridge details is spent Figure 97. Comparison of crack growth rates for structural steels initiating fatigue cracks and growing them at values of AK lower in air and aqueous environments. [Yazdani 1983] than those at which corrosion fatigue crack growth rates were measured in most previous tests. To determine the effect of corrosion on the total fatigue life, it is necessary to also examine the available CFCI data and the available CFS-N data. in sodium chloride solution, despite strong attempts to characterize the long-life behavior of primary interest for structural applications such as bridges. The long-life (3,000,000 cycles) cyclic strength for CFCI Corrosion Fatigue Crack Initiation behavior was = 34 ksi (234 MPa) for all steels. The degradation in cyclic load-carrying capacity, as de- Novak determined the CFCI life of A36, A588 Grade A, and termined by comparing the FCI and CFCI data, varied from a A517 Grade F steel specimens tested in a 3.5 percent sodium negligible amount at 1,000 cycles of loading to maximum chloride solution under constant-amplitude cyclic loading and amounts of about 54, 62, and 72 percent, respectively, for the a realistic cyclic frequency of 12 cpm [Novak 1983b]. The CFCI A36, A588 Grade A, and A514 Grade F steels at 3,000,000 life was defined as the number of cycles needed to initiate a cycles. - crack from a notch with varying root radius (stress concentra- Taylor and Barsom had reported similar findings for CFCI tion) and to grow it to a surface length of between 0.030 and in A5 17 Grade F steel specimens [Taylor 1981]. 0.070 in. (0.75 and 1.75 mm). The data were presented in plots The data show that the CFCI life of weathering steel details of CFCI life versus the maximum stress range at the notch tip. in an aqueous sodium chloride environment is much shorter The latter is equal to the nominal stress range times the theo- than the fatigue crack initiation life in a clean air environment. retical stress concentration factor, A07ma. = 10,om K An eval- There appears to be no fatigue limit in this aqueous environment. uation of the data showed that [Novak 1983b]: Therefore, the fatigue strength of long-life structures, such as bridges, in a corrosive environment may be much lower than The fatigue crack initiation (FCI) thresholds in air were the AASHTO fatigue limits for "over two million cycles" of = 73, 90, and 124 ksi (505, 623, and 855 MPa), respec- loading which were based on the results of fatigue tests per- tively, for A36, A588 Grade A, and A5 17 Grade F steels. formed in clean laboratory air and intended for painted struc- No CFCI thresholds were found for any of the steels tested tures [Fisher 1970, Fisher 1974]. 116

Corrosion Fatigue S-N Data Weathering fatigue life (Chapter Seven and Fig. 94). Corrosion fatigue crack initiation life (Chapter eight; sec- A further indication of how a corrosive environment might tion under "Corrosion Fatigue"). affeót the total fatigue life (crack initiation plus propagation) of Crack growth rates (Figs. 95, 96, 97). weathering steel details can be obtained by comparing the fatigue Corrosion fatigue S-N data (Fig. 98). lives of specimens tested to failure in air and aqueous environments. Such S-N tests include crack initiation and crack prop- Items 1 and 2 affect the crack initiation life, item 3 the crack agation at all values of AK, from near threshold to failure. propagation life, and item 4 the total life. These effects are not Knight collected such data from the literature [Knight 1977]. necessarily additive. The cyclic history of bridges is a complex Table 46 summarizes the data and the calculations performed phenomenon not easily modeled in short-term laboratory tests. herein. The fatigue notch factors were obtained by dividing the Weathering steel bridges are subjected, at first, to weathering stress range in air at 2,000,000 cycles into the stress range for fatigue during construction and early years of service. If the the Category A mean line at 2,000,000 cycles. The latter was stress cycles are sufficiently high to initiate fatigue cracks from calculated from Table 39: weathering-induced rust pits and/or by corrosion fatigue crack initiation, the phenomenon gradually changes into one of cor- = (1013785/2,000,000)U3178 = 32.8 ksi (226 MPa). rosion fatigue crack propagation. One can estimate the total corrosion fatigue life of a weathering steel bridge in two ways. In one, which parallels the current The loss in life was calculated from Eq. 32, using the stress AASHTO approach of defming the life based on fatigue tests ranges in air and in aqueous environments reported by Knight. performed in air, one would repeat such tests in an aqueous The results, plotted in Figure 98, suggest the following conclu- environment. Figure 98 summarizes such early corrosion fatigue sions: S-N data that were obtained from tests of specimens with Cat- egory A and B strength. Excluding the stress reversal data (points 5 and 10), the More recent data on welded joints with fatigue strengths loss in life due to corrosion fatigue varied from 55 percent to comparable to those of Category C to E details were developed 97 percent. during extensive European research programs on offshore struc- The loss in corrosion fatigue life did not significantly de- tures. As a result of such studies, the new British design rules crease with increasing fatigue notch factor, as was observed in for offshore structures now specify a 50 percent reduction in the weathering fatigue tests (Fig. 94). total fatigue life for all unprotected joints in sea water, that is Salt-water environments reduced the corrosion fatigue life [UK DOE 1981]: more than fresh-water environments. Type of steel was not a significant variable. AN, = 0.50 (43) As an example, Figure 99 shows the S-N data for welded cruciform joints fabricated from SM 50 structural steel corre- Equation 43 is plotted in Figure 98. The same reduction for sponding to point 11 in Figure 98 (Serizawa 1973). In thiscase, corrosion fatigue should also apply to bridges that are subjected salt-water immersion decreased the fatigue life by 84 percent at to intermittent corrosion, because the crack creates its own 2,000,000 cycles. It also drastically lowered the fatigue limit environment at the tip, with a chemistry different from that of from about 18 ksi (123 MPa) in air to 7.5 ksi (52 MPa) in salt the bulk environment outside of the crack. In fact, differences water, a finding consistent with the data for CFCI life [Novak in bulk environment were shown to have no significant effect 1983b, Taylor 1981]. on crack growth rate fatigue life as, for example, in the following steel-environment combinations:

Summary Crack growth rates for BS 4360 Grade 500 structural steel in natural seawater, half diluted seawater, seawater with All available data show large reductions in fatigue life of 2 mg/l HCO3, and 3.5 percent NaCl [Scott 1975]. structural details in aqueous environments. This conclusion ap- Crack growth rates for A36, A588, and A514 steels in plies to all structural steels, not just weathering steel. No data distilled water and 3 percent NaCl solution, Table 45 [Barsom reported in the literature show a beneficial effect of aqueous 1977]. environments. It can be expected, therefore, that in corrosive Corrosion fatigue (S-N) line for BS 4360 Grade 500 steel environments weathering steel bridges will have a shorter fatigue in seawater under freely corroding and intermittently corroding life than painted bridges. conditions [Booth 1983]. 5083-Hl16 aluminum, HY-80 steel, and 17-4 PH H1075 steel in natural seawater, ASTM seawater, and 3.5 percent NaCl ESTIMATION OF LOSS IN LIFE solution [Bogar 1983].

Estimates of the total fatigue life of weathering steel bridges Additional examples are given in Figure 98 and in the ref- should consider the effect of weathering fatigue on the crack erence [Burnside 1983]. initiation life, and the effect of corrosion fatigue on the crack Furthermore, the capillaries in the oxide coating and the crack initiation and crack propagation lives. The available data are of can retain humidity for long periods of time after precipitation the following types: or condensation has stopped. This humidity, in conjunction with 117

Table 46. Summary of corrosion fatigue data for specimens tested in both air and aqueous environments.

Stress Range at 2,000,000 Cycles Fatigue Point in Steel & Tensile Type of Aqueous In Air In Aqueous Loss in Notch Figure 98 Strength Detail Environment Environment Life Factor K (MPa) (MPa) (MPa)

1 785 Base Salt water 430 235 86 0.53 metal immersion

2 785 Base Dripping 460 150 97 0.49 metal salt water

3 785 Base Dripping 320 225 68 0.71 metal fresh water

4 490 Base Dripping 300 160 87 0.76 metal salt water 5a 0.35% C Base Synthetic 454 423 20 0.50 steel metal tap water

6 785 Butt weld Salt water 180 110 79 1.26 manual arc immersion

7 785 Butt weld Dripping 167 130 55 1.36 manual arc fresh water

8 785 Butt weld Dripping 223 165 62 1,02 submerged fresh water arc

9 785 Butt weld Dripping 266 192 65 0.85 manual arc fresh water toes ground

10a Mild steel Butt weld Dripping 200 162 49 1.13 410 salt water

11 490 Welded Salt water 153 87 84 1.48 cruciform immersion

12 Low-alloy Welded Dripping 315 185 82 0.72 780 tee" salt water

Notes

a. R = -1; all others R = 0.

chlorides and other corrosive contaminants trapped in the oxide Albrecht 1983b]. From the data plots of Barsom [1977], Al- coating, fosters corrosion beyond the time of surface wetness. brecht had estimated that aqueous environments increase the It is worth remembering that immersion in a 3 percent sodium crack growth rate of A588 steel in air by a factor of VT = chloride solution has been used in a past NCHRP study [Barsom 1.41. Recent analysis of the data, presented earlier in this chapter 1977] and a current FHWA study [Roberts 1983] as a standard under "Corrosion Fatigue," showed that the factor was 1.48 at environment for determining the corrosion fatigue behavior of the lowest iK values tested and more at higher i.K values bridge steels. The water runoff from highway bridges can have [Yazdani 1983]. a NaCl concentration higher than the 3.5 percent NaCl con- The relative loss in life due to weathering is given by Eqs. 34 centration in seawater. Therefore, the previously mentioned data and 35. The relative loss in life due to corrosion fatigue is, and Eq. 43 are also indicative of the loss in life due to corrosion therefore, 0.41/1.41 = 29 percent of the relative net life after fatigue that can be expected for all structural details in weath- weathering. For K1 2.63: ering steel bridges. Another method of estimating the total corrosion fatigue life = 0.29 (1 —Mc) = 0.11 K (44) of a weathering steel bridge was proposed by Albrecht. It as- sumes that weathering fatigue consumes the crack initiation life and, for K1 > 2.63: and corrosion fatigue reduces the crack propagation life, that is the net life after the loss due to weathering [Albrecht 1982a, = 0.29 (45) 118

E 5 E 2 ui 4I 0< a ,< _CD Cl) E a o e E oE o g 'E

U C.) c.,i— Qll)- C.)— 01.)

0 Salt Water II (1 Fresh Water 08 2 A 10 60 A!brecht 1982n

Weathering and 0 - 7,L Corrosion Fatigue

UK. DOEf ,—Eq. 46, 0. Eq. A3 ,çEci4? • — Eq. 4 - - 0. 0 — - -- -- Corrosion Fatigue I i I i I 0 I 2 3 4

Figure 98. Loss in life due to corrosion fatigue. FATIGUE NOTCH FACTOR, Kj

0 —40

200 - Air a-0 S

Ui IOO- z 4 Salt Water 4

CI) CI) C', 10 U, Ui Ui I— 50 U, U Salt Water Category D

Ii i I I I I Figure 99. Corrosion fatigue strength 2 5 2 5 10 5 o6 of SM 50 welded crucjform joints im- • CYCLES TO FAILURE mersed in salt water. [Serizawa 1973]

Equations 44 and 45 for loss in CFCP life are plotted with = 0.29 (47) a dashed line in Figure 98. The loss predicted from crack growth rates greatly underestimates the loss in total corrosion fatigue Equations 46 and 47, originally developed in 1981, were up- life (crack initiation and propagation) that was determined from per-bound estimates of the loss in life due to weathering and S-N data. Adding Eqs. 45 and 46 to Eqs. 34 and 35, respectively, corrosion fatigue for data available at that time. Including the gives the total relative loss in life due to weathering and corrosion more recent weathering fatigue data and the corrosion fatigue fatigue. For K < 2.63: data, respectively, Figures 94 and 98 show that Eq. 46 is no longer an upper-bound estimate of the loss in life due to weath- AN, = + Mr1 = 1 - 0.27 K1 (46) ering and corrosion fatigue. Equation 46 is nonconservative. It underestimates the experimentally determined losses. and, for K1 > 2.63: Substituting the fatigue notch factors from Table 39 into Eq.

119

46 gives the following predicted losses in life: 73 percent for ering and corrosion fatigue, given by Eqs. 46 and 47. On that Category A, 60 percent for Category B, 36 percent for Category basis, the allowable stress ranges for weathering steel bridges, C, and 29 percent for Categories D to E prime details. These F,,,, are equal to Eq. 52 multiplied by the relative net stress values are the intersection points of Eq. 46 with the vertical ranges, given by Eqs. 50 and 51. For K1 < 2.63: grid lines in Figure 98. V Subtracting Eqs. 46 and 47 from unity yields the total relative l01_2 F,,, = (0.27 K1 (53) net life after weathering and corrosion fatigue. For K1 < 2.63: Nd

N, = 1 - N, = 0.27 K1 (48) and, for K1 > 2.63:

and, for K1 > 2.63: l0L2 V'n = (0.71 (54) N, = 0.71 (49) The same reductions were also applied to the loading condition "for over 2,000,000 cycles." This seems reasonable, since Novak Since fatigue life is an m-th power exponential function of stress found a very large degradation in the cyclic load-carrying ca- range, the corresponding total relative net stress ranges are, for pacity at 3,000,000 cycles for all structural steels tested in a K1 < 2.63: sodium chloride solution [Novak 1983b]. Other investigators have also reported a degradation of the fatigue limit in corrosive = (0.27 K)"" (50) environments. Substituting the values of b, s, m, and K1 from Table 39 into and, for K1> 2.63: Eqs. 53 and 54 yields the allowable stress ranges for redundant- load-path structures after losses due to weathering and corrosion = (0.71)" (51) fatigue. The results are shown in Table 47(b). The allowable stress ranges for nonredundant load path structures, given in In summary, Eq. 43 or Eqs. 46 and 47 estimate the total loss Table 48(b), were obtained in like manner. in life of weathering steel bridges. The former is nonconservative Figure 100 compares the AASHTO allowable stress ranges for Category A to B details. The latter better fits the data for at 500,000 cycles (open bars) with those proposed for weathering Category A and B details, but is nonconservative for the Cat- steel bridges after total losses due to weathering and corrosion egory C to E details. fatigue (solid bars). As can be seen, the reductions vary from about one and one-half category at A to one-half category at E. For information, the cross-hatched bars show the allowable FATIGUE DESIGN STRESSES stress ranges after the loss due to weathering alone. The results

Design Equations are the same for the other AASHTO loading conditions of 100,000 and 2,000,000 cycles. The only fatigue design equations proposed to date for weath- An alternative approach to that proposed in Albrecht [1982a] ering steel bridges are those given in Albrecht [1982a] and would be to apply to all structural details a uniform reduction Albrecht [1983b]. Those equations and an alternative approach in life, such as the 50 percent adopted by the U.K. Department are presented in the following. of Energy for offshore structures [UK DOE 1981]. In this case, The allowable stress ranges, F,,, in the AASHTO fatigue the allowable stress ranges would be given by combining Eqs. specifications were obtained by shifting the mean S-N lines to 43 and 52. the left by a distance equal to two standard deviations of the 0.5010 b— 2V'"V. log of life. This gives: F , = (55) ( Nd) b V' F,,= (10-2 (52) -- Assuming the average slope of all S-N lines to be m = 3.2 (Table 39) yields a relative net stress range off,, = 0.501132 = where Nd equals the number of design cycles, b equals the 0.80, that is a 20 percent loss in stress range. The corresponding intercept, m equals the slope, and s equals the standard devia- allowable stress range levels are marked with arrows in Figure tion. The data base comes from fatigue tests performed in clear 100. In comparison with the equations proposed by Albrecht, air [Fisher 1970, Fisher 1974]. The regression coefficients and Eq. 55 gives higher allowable stress ranges for Category A and the standard deviation for the Category A to E' data are sum- B details and lower stress ranges for Category C to E' details. marized in Table 39. Substituting them into Eq. 52 gives the The uniform reduction approach is attractive for its simplicity. allowable stress ranges for redundant-load path structures given in Table 47(a). The differences between those values and the ones listed in the AASHTO specifications arise from the round- Comments on Design Equations off errors of 4 percent, at most, in the latter [Albrecht 1981]. To arrive at design stress ranges for weathering steel bridges, The approach that was followed in deriving the fatigue design Albrecht shifted the current AASHTO design lines for clean equations (Eqs. 53 and 54) for bare weathering steel structures air by an amount corresponding to the loss in life due to weath- consisted of shifting the AASHTO design lines (lower confidence

120

Table 47. Allowable stress ranges for redundant load path structures.

Allowable Stress Range Category

For 100,000 Cycles For 500,000 Cycles For2,000,000 Cycles For Over 2,000,000 Cycles ksi MPa ksi MPa ksi MPa ksi MPa

(a) Current AASHTO Specifications (Eq. 52)

A 61.2 422 36.9 254 23.9 164 24 165.47 B 45.0 311 27.9 193 18.5 128 16 11031 C* 34.7 239 20.6 142 13.2 90.8 12 82.74 C 32.5 224 19.8 136 12.9 89.1 10 68.95 D 26.8 185 15.9 110 10.1 69.7 7 48.26 E 21.0 145 12.5 86.0 7.97 54.9 5 34.47 E 15.2 104 9.17 63.2 5.94 41.0 2.6 17.93

(b) After Weathering and Corrosion Fatigue (Eqs. 53 and 54)

A 40.6 280 24.4 169 15.9 109 15.9 109 B 34.4 237 21.3 147 14.1 97.4 12.2 84.2 C* 28.2 195 16.8 116 10.7 73.8 9.75 67.3 C 28.2 195 17.2 119 11.2 77.4 8.69 59.9 D 24.0 165 14.2 98.0 9.05 62.4 6.26 43.2 E 18.8 129 11.2 77.0 7.13 49.2 4.48 30.9 E 13.6 93.9 8.23 56.8 5.34 36.8 2.34 16.1

* For transverse stiffener welds on girder web and flanges.

Table 48. Allowable stress range for nonredundant load path structures.

Allowable Stress Range Category

For 100,000 Cycles For 500,000 Cycles For 2,000,000 Cycles For Over 2,000,000 Cycles ksi MPa ksi MPa ksi MPa ksi MPa

(a) Current AASHTO Specifications (See Table 1.7.2A1)

A 36 248.21 24 165.47 24 165.47 24 165.47

B 27.5 189.60 18 124.10 16 110.31 16 110.31

C 19 131.00 13 89.63 10, 12* 68.95, 82.74* 9,11* 62.05, 75.84*

O 16 110.31 10 68.95 7 48.26 5 34.47

E 12.5 86.18 8 55.15 5 34.47 2.5 17.24

(b) After Weathering and Corrosion Fatigue

A 24.4 169 15.8 109 15.9 109 15.9 109

B 21.3 147 14.1 97.4 12.2 84.2 12.2 84.2

C* 16.8 116 10.7 73.8 9.75 67.3 8.94 61.7

C 17.2 119 11.2 77.4 8.69 59.9 7.82 53.9

o 14.2 98.0 9.05 62.4 6.26 43.2 4.48 30.9

E 11.2 77.0 7.13 49.2 4.48 30.9 2.25 15.5

* For transverse stiffener welds on girder web and flanges. 121

N. , ".

- 30 200

100 E U) - LUCr I- ice E U, Figure 100. Comparison of allowable stress ranges for redundant loadpath structures at 500,000 cycles: (1) current AASHTO, in clean air; (2) after weath-

iii 111111111 111111 ering; and (3) after weathering and cor- F 0 D E rosion fatigue.

limits) to account for the loss in fatigue life caused by both the or. ideally exposed for 4 years. As a result, the following factors effect of weathering prior to load cycling and the effect 'of might tend to cause Eqs. 53 and 54 to overestimate fatigue life exposure to an aqueous environment during load cycling. These for in-service bridge members. effects are additive because the former reduces the crack initiation life, while the latter reduces the crack propagation life. Long-term loss in fatigue life caused by 50 or more years In developing Eqs. 53 and 54, the standard deviation of the of weathering should be expected to exceed the loss after 4- S-N data for weathered specimens was set equal to that for years exposure. In fact, recent data, for 8-year-weathered, au- nonweathered specimens. This is a generally conservative as- tomatically welded stiffeners show that their mean loss in life sumption and is intended to compensate for several other factors (point 15 in Fig. 94) exceeds the value predicted by Eq. 34 that are not taken into account directly in these equations. (dashed line in Fig. 94). The results of the regression analysis, given in Tables 40 and The corrosion penetration data presented in Chapters Four 41, showed that the standard deviation after the initial 2 to 3 and Five show that the corrosion and pitting of structures in years of weathering: service is more severe than the corrosion of ideally and boldly exposed specimens. Decreased for Category A and B details. Loss of net section due to corrosion increases the stress Did not change for Category C details. range. Increased for the details whose mean fatigue life increased Reported losses in crack propagation life (Fig. 97) caused with weathering. by corrosion fatigue in an aqueous (as opposed to air) environment are larger than the assumed 29 percent (Eq. 45 and Fig. Weathering times of 4 to 8 years did not change the standard 98). deviation when compared with the values calculated after the Results of-the regression analysis, given in Table 45, show initial 2 to 3 years. The following examples illustrate how the that the standard deviation of crack-growth-rate data is larger changes in the standard deviation affected the lower confidence in an aqueous environment than in an air environment. Ac- limit (L.C.L. = mean minus two standard deviations): cordingly, the loss in corrosion fatigue life would actually be greater than that predicted from the loss in mean corrosion Figures 79 and 80: The L.C.L. for weathered base metal fatigue life. fell below the AASHTO allowable line for Category A. Figure 86: The L.C.L. for weathered transverse stiffeners In view of these factors, it does not appear to be sufficient fell on, or slightly above, the AASHTO allowable line for to compare the lower confidence limit of the available weath- Category C. ering fatigue data against the AASHTO allowable S-N lines and recommend a one-category penalty on the fatigue resistance of It is worth noting that the L.C.L. continued to degrade with Category A details fabricated from weathering steel, and no exposure times longer than 3 to 4 years (Fig. 86). penalty on the fatigue resistance of Category C to E' details Data used to establish the mean loss in fatigue life for weath- [Fisher 1983b]. The proposed design Eqs. 53 and 54 attempt to ering steel are based on dry-air fatigue tests on specimens boldly anticipate the long-term 50-year losses in fatigue life, the in- 122

creased corrosion under nonideal exposure conditions, the loss Finally, Yamada and Albrecht reported fatigue test data for in net section, and the detrimental effect of aqueous environ- splices of plate girder flanges with 1:4 thickness transition [Ya- ments. mada 1977]. All data points fell above the Category B allowable With the information available in 1981, when the reference line. The 13 details failed from cracks that initiated at external [Albrecht 1 982a] was issued, it was believed that the conserv- flaws. ative and nonconservative assumptions would compensate each In summary, most (196 of 250) Category B details failed from other and would not lead to underestimates of the 50-year loss external flaws. Since weathering and corrosion fatigue affect in life of weathering steel bridges. In retrospect, this expectation crack growth from external flaws, the loss in life must be applied may not be realistic. to Category B details.

Category B Details Other Viewpoints Some investigators stated that the Category B allowable line In the opinion of some investigators, the allowable stress range need not be lowered because most fatigue cracks at Category B need not be reduced so long as the weathering fatigue data for details initiate at internal, subsurface flaws that are not affected a detail do not fall below the applicable AASHTO allowable by weathering and corrosion fatigue on the surface [Fisher line [Fisher 1983b, Barsom 1983]. On that basis, Fisher agreed 1983b, Barsom 1983]. to a one-category reduction in fatigue strength for base metal, In reality, most Category B details fail from external flaws, from A to B, but not for any other detail. Barsom concluded as the following data show [Albrecht 1983c]: that the AASHTO allowable lines, including that for Category A, apply equally to weathered and nonweathered steel members. In their opinion, rust pitting provides a notch condition not NUMBER OF FAILURE more severe than the worst flaw in a given detail. For this reason, INITIATIONS they added, rust pitting lowers the mean life but not the lower REFERENCE TYPE OF DETAIL INTERNAL EXTERNAL confidence limit. Both Fisher and Barsom are of the opinion that the detri- Fisher 1970 Welded beams 29 21 mental effect of corrosion fatigue need not be accounted for in Fisher 1970 Flange splices with 25 137 width transition designing weathering steel bridges. Yamada 1977 Flange splices with 0 13 thickness transition Other Specifications

The Category B allowable line was based on the fatigue test The domestic and foreign specifications for fatigue design of data developed by Fisher, et al., for welded beams [Fisher 1970]. aerospace structures, nuclear reactors, pipes, boilers, machinery, Hirt subsequently examined the fatigue failures of the 50 welded pipelines, and offshore structures provide for reduced fatigue beams from the reference [Fisher 1970] and reported that the strength when the structure is not protected against corrosion. cracks initiated from internal porosities in 29 beams and from Although the first weathering steel bridge in the United States external discontinuities in 21 beams [Hirt 1971]. The two sets was built in 1964, the AASHTO, AREA, AISC, and AWS of welded beam data were evaluated herein with regression and specifications do not have provisions for fatigue design of non- statistical analyses. The regression line, standard deviation, and painted steel structures. fatigue notch factor are: (1) for beams failing from internal European code writing bodies are more aware of the need to porosities: log N = 12.8553 - 2.984 logf, (MPa), s = 0.07 17, consider the detrimental effect of corrosion on fatigue than their K = 1.40; and (2) for beams failing from external discontinuites: American counterparts. In comparing the British and American log N = 13.6820 - 3.394 logJ (MPa), s = 0.01350, K1 = fatigue design rules for welded structures, Gurney and Maddox 1.56. The 10 percent difference in fatigue strength of the two stated that [Gurney 1980]: "Allowance for environment may sets was insignificant at the 5 percent level of significance, as also lead the reductions in allowable stresses, including the determined by a t-test of the differences between the intercepts fatigue limit. In this context it may be noted that the British and the slopes of the two regression lines. It follows that Cat- rules for the design of fixed offshore structures (DD55) allow egory B welded beams are statistically as likely to fail from for seawater corrosion by reducing the fatigue limit." external as from internal discontinuities. Current drafts of European codes and recommendations for In a second series of tests on Category B details, Fisher, et fatigue design of steel structures state that "reduction in fatigue al., obtained fatigue data for 162 welded beam flange splices life due to corrosion" is not explicitly covered [Eurocode 1983, with either straight or 2-ft (600-mm) radial transition in width ECCS 1982]. The ECCS recommendations limit the application [Fisher 1970]. Twenty-five of the 162 butt-welded splices failed of the fatigue strength curves to structures in normal atmos- from gas pockets, oxide inclusions, and internal cavities. The pheric conditions with suitable corrosion protection. They fur- majority of the splices (137) failed from external flaws. The ther state that more corrosive environments may significantly straight and curved transitions had fatigue notch factors of 1.47 reduce the fatigue strength. and 1.35, respectively; hence, they had slightly longer mean lives A Specialist Drafting Panel recommended revision of the than the plain welded beams for which IC = 1.49. Both are fatigue sections of the U.K. Department of Energy's document Category B details, and most beams failed from external dis- "Offshore Installations: Guidance on Design and Construction" continuities. [UK DOE 1978, Tomkins 1982]. The revisions include an across 123

the board factor-of-two penalty on life (50 percent loss in life) "for 2,000,000 cycles" of truck loading on two or more traffic for unprotected steel welded joints exposed to seawater. Pro- lanes, for which the wheel load distribution factor is S/5.5. tection is taken to include: (1) paint and other coatings, where Category D, E, and E' details are always governed by the loading these are known to adequately prevent contact of water with condition "for over 2,000,000 cycles" of truck loading on one the steel; and (2) electrochemical potential. Furthermore, no traffic lane, with distribution factor S/7 [Albrecht 1982b]. fatigue limit is assumed to exist. The S-N design curve for seawater corrosion fatigue falls, therefore, a factor-of-two on life below the basic air curve (T) and extends as a straight line Example No. 1 to 100 million cycles, at a slope of —1/3. - In the Federal Republic of Germany, special permits for Eight simple-span bridges; noncomposite, rolled beams; HS- building highway bridges with orthotropic steel decks and rail- 20 loading; 25-ft to 60-ft (7.62-rn to 18.30-rn) spans; 30-ft (9.15- way bridges from weathering steel are not granted because the m) roadway; 8-ft 7-in. (2.62-ni) girder èpacing; 7-in. (0.1 8-m) authorities are concerned about the detrimental effect' of cor- concrete deck; A588 steel, F = 50 ksi (345 MPa); 4,000-psi rosion on fatigue life. (27.6-MPa) concrete; interior girders; working stress design The available data show that steel structures not protected [USS 1975a]. against corrosion have shorter fatigue lives. Therefore, allowable The stress ranges were computed for two details, both at fatigue stress ranges for weathering steel in aqueous environ- midspan: (1) Category , A base metal, extreme fiber of tension ments should be lowered from their present levels for clean air flange; and (2) Category C diaphragm gusset welded to the web, environments. symmetrically about the centroid of the girder. The gussets were at most 9 in. (230 mm) long. The calculated midspan stress ranges, shown in Figure 101 APPLICATION TO CURRENT DESIGNS with triangular symbols, are much lower than the 24 ksi (165 MPa) allowed by AASHTO for Category A base metal, but It is of interest to examine the safety margin against fatigue they exceed the proposed 15.9 ksi (109 MPa) allowable stress of highway bridges designed to the current AASHTO specifi- range for weathering. steel bridges. Fatigue would control the cations, in terms of the lower allowable stress ranges proposed flexural design of this set of bridges. The Category C diaphragm previously under "Fatigue Design Stresses." These series of gussets are in no case fatigue critical, simple-span A588 steel bridges were investigated [Albrecht 1982a]. The following summarizes the types of bridges and the cal- Example No. 2 culated stress ranges at the fatigue critical details. The latter are then compared in Figure 101 with the: (1) stress ranges Eight simple-span bridges; composite rolled beams; HS-20 allowed by AASHTO for redundant load path structures; and loading; 25-ft to 60-ft (7.62-m to 18.30-rn) span; 30-ft (9. 15-m) (2) values proposed in Table 47 for weathering steel structures. roadway; 8-ft 7-in. (2.62-rn) girder spacing; 7-in. (0.18-rn) con- All bridges were designed for an average daily truck traffic, crete deck; A588 steel, F1 = 50 ksi (345 MPa); 4,000-psi (27.6- ADIT = 2,500. In this case, the fatigue design of Category A, MPa) concrete; interior girders; working stress design; unshored B, and C details is always governed by the loading condition construction [USS 1975a].

25 HS-20, COMPOSITE, REF CuSS 19773 24 AASHTO , HS - 20, NON - COMPOSITE, REF (USS1977] HS-l5, COMPOSITE, With COVER PLATE, REF ESIMON 19811

20 20 Clean Air, AASHTO - -- Weathering And Corrosion Fatigue IS

15 15 15 I Proposed AASHTO-7 Ui 13 z 4 Prposed 10 10 10 10 (I) U, Ui I- U, AASHTO--1ASHT0- 5 I 5 oo'I Figure 101. Comparison of calculated CAT.A CAT. C 020 ' CAT.E stress ranges in simple-span bridges 0 with allowable stress ranges for: (1) 0 . 0 20 40 60 20 40 60 20 40 60 40 60 clean dir, current AASHTO; (2) weath- SPAN LENGTH (f t) ering and corrosion fatigue. 124

As in the previous example, the midspan stress ranges were for Category Band C. But they exceed the proposed stress range computed for Category A base metal and Category C diaphragm for Category E by 0.5 ksi (3.6 MPa). gussets. The calculated stress ranges are plotted inFigure 101 with circular symbols. The composite section raised the neutral Comparison with Measured Stress Ranges axis and substantially increased the stress range at the lower end of the diaphragm gusset. It moderately lowered the stress Another way of assessing the impact of lowering the allowable range at the extreme fiber. Fatigue only controls the design of stress ranges on bridges already in service is to cornpare the both details when the spans are short. highest stress range measured on bridge girders against those proposed herein for weathering steel bridges. Shaaban and Al- brecht collected and analyzed 190 stress-range histograrns that had been recorded on 40 bridges in Michigan, Alabama, Mary- land, Virginia, Tennessee, Minnesota, Connecticut, Illinois, and Example No. 3 Ohio [Shaaban 1983]. Table 49 compares the results for the most highly stressed detail in each category. Four simple-span bridges; composite rolled beams with cover In Category A and B details, for which the observed reduc- plates; HS-15 loading; 45.ft to 60-ft (13.72-rn to 18.30-rn) span; tions in fatigue strength are greatest, the highest measured stress 24-ft. (7.32-rn) roadway; 6-ft 9-in. (2.06-rn) girder spacing; 7- ranges are lower than the proposed allowable stress ranges and in. (0.18-rn) concrete deck; A588 steel, F = 50 ksi (345 MPa); well below the fatigue lirnits. In these cases, weathering is not 4,000-psi (27.6-MPa) concrete; interior girders; working stress expected to reduce the fatigue life of existing bridges. Stress- design; unshored construction [Simon 1981]. range histograms were not recorded at Category C and D details. The stress ranges were computed for three details; (1) Cat- In category E and E' details, for which the proposed reduc- egory B cover plate-to-flange weld at midspan; (2) Category C tions are small, the highest measured stress ranges significantly diaphragm gusset at midspan, max. 9-in. (230 mrn) long; and exceed the proposed as well as the current AASHTO allowable (3) Category E cover plate end. stress ranges. In these cases, heavy truck loading, more than The calculated stress ranges, plotted in Figure 101 with rec- weathering, may shorten the fatigue life of bridges below the. tangular syrnbols, are much lower than the proposed allowables values intended by the AASHTO specifications.

Table 49. Comparison of allowable, computed, and measured stress ranges in highway bridges. [Shaaban 1983]

Stress Type of Allowable Stress Range Computed Highest Category Detail Bridge Current Proposed Stress Measured AASHTO Range Stress (ksi) (ksi) for HS20 Range (ku) (ksi)

A Base metal Shaffer Creek, F.A.I. 24.0 15.9 16.3 8.93 at midspan Rt 74, Illinois

B Longitudinal Overpass for Southern 18.0 14.1 9.20 7.75 flanye-to RR, Sayre to Alden cover plate weld Rd., Jefferson County, at midspan Alabama C------No data available------D------No data available------

E Flange at cover Over Alabama River on 5.0 4.5 6.20 7.25 plate end Road from Hunter Station to Prattmont, Montgomery and Elmore Counties, Alabama

Flange at cover Overpass for Southern 2.5 2.2 14.10 7 .25 plate end RR. Sayre to Alden Rd., Jefferson County, Alabama 125

CHAPTER NINE

PAINTING

This chapter deals with three aspects of painting. First, the petrochemical plant, the vinyl coating was severely undercut on performance of paint systems applied to a fresh weathering steel the carbon steel and slightly undercut on the weathering steel substrate are compared to the same paint systems applied to specimens. mild steel. Second, the Japanese practice of treating weathering steel with a sacrificial rust stabilizing system is described. Third, Wonders evaluated the performance of paint on carbon and the results of laboratory tests of remedial paint systems for weathering type steels [Coburn 1978]. Standard specimens of severely corroded weathering steel specimens are reviewed. Ex- both steels were exposed for several months in a coal coking amples are given of paint systems that were used to rehabilitate plant atmosphere similar to those found in steel plants world- weathering steel bridges. wide. In this type of environment, wind-blown moisture droplets carrying ámonium chloride deposit on metallic structures. They cause accelerated deterioration of the paint and attack the un- INITIAL PAINTING OF NEW BRIDGES derlying steel. After the initial exposure, some specimens were sandblasted and others were hand wire brushed. The specimens Several investigators have examined the performance of paint were then coated with organic or inorganic zinc-rich alkyd, systems applied to a nonweathered or fresh substrate of weath- epoxy, or vinyl primer, followed by one of seven generic types ering copper (copper-bearing) and carbon steels. Some have of intermediate or top coat systems. All painted specimens were found certain paint systems to be more durable on a weathering inscribed (X) down to the base metal and weathered for one steel substrate [Copson 1959, Schmitt 1967, and Coburn 1978]. year on racks facing the source of the pollutant at a 30-deg Others have found comparable durability of the same paint angle from the horizontal. The following results were obtained: systems applied to both weathering and carbon steels [Nakay- ama 1972, Van Eijnsbergen 1979, and Scantlebury 1980]. The So long as both the organic or inorganic zinc-rich primers results of the aforementioned studies are reviewed in the were in direct contact with the steel surface, regardless of the following. top coat, all specimens, were in excellent condition. Copson and Larrabee examined the durability of paint on The inorganic zinc-rich primer outperformed the organic sheets of copper and HSLA steels on the sides of a railroad zinc-rich primer when both were top coated with the same epoxy hopper car [Copson 1959]. After 5-years service, the paint on or vinyl paint. the copper-bearing steel sheets had come off in relatively large sheets, whereas the paint on the HSLA steel sheets continued For this series of test, Coburn [1978] did not assess the relative to give adequate protection. Because the rust that formed on durability of paint on carbon and weathering steels, or the effect the HSLA steel is less voluminous, the paint film ruptured to of initial exposure and cleaning method on paint durability. a lesser extent and, thus, less moisture reached the steel to Another series of one-year exposure tests examined the per- promote corrosion. formance of 48 combinations of protective coating systems and In the same study in which the performance of bare carbon, substrates [Coburn 1978]. Table 50 summarizes the results. A A242 Type 1, and A588 Grade A steels in chemical plant rating of 10 indicates no rusting, while a rating of zero implies environments was evaluated (Chapter Five, section under "In- complete rusting. The data show that the paint systems con- dustrial Pollutants"), Schmitt and Mathay also examined the sisting of an organic zinc-rich primer with or without a vinyl durability of a paint system applied to these steels [Schmitt top coat, or an inorganic zinc-rich primer with or without an 1967]. The steel specimens were sandblasted, spray coated with epoxy top coat were about equally durable on both the carbon a wash primer, and top coated with a 4-mils to 5-mils (100-m and weathering steels. However, the paint systems that did not to 125-rn) thick dry film of vinyl commonly used for chemical adequately protect the steel against corrosion lasted longer on plant structures. A mark (X) was inscribed on the skyward the weathering steel than on the carbon steel, when both were surface of the coated specimens. The specimens were exposed exposed one year. The reason for this is that the rust on exposed at most 2 years in the environments listed in Table 35. The areas, such as at scratches in the paint on the weathering steel, performance of the vinyl coating was judged by the extent of is less voluminous and, therefore, undercuts the paint film to a undercutting at the inscribed mark. The findings were as follows: lesser degree. When exposed to severe environments that cause rapid paint For exposure periods up to one year, the vinyl coating on failure, such as continuous dampness and chlorides, the under- the weathering steels tended to be less undercut than on the lying weathering steels also corroded and produced flaky and carbon steel. voluminous rust comparable to that found on carbon steel. Un- After 2-years exposure at one sulfur and two chlor-alkali der such conditions, there is no reason to expect that the paint plants, the vinyl coating completely failed on all steels. In a on the two steels would behave differently. Table 51 shows the 126

Table 50. Relative performance of various coating systems on carbon and weathering steels in two coke plants. [Coburn 1978]

S St

5.5;'

I RO 20

Iii og

• P. ZO RIF ______• II" I g I ' I d h i 00 MMMMMMMM~ UJIi.T4

LTTI - —IIL— —LL1Ij

Table 51. Relative performance of various coating systems on carbon and weathering steels in Kure Beach (the 24-rn lot), N.C. [Coburn 1978]

S. C Carbo n

5; W A242 Type

,0070 F ;'0 120 go

20' lzi P R I Flo

;011

I LF LE 11 127 results of two additional paint durability tests performed at Kure slight difference in the amount of paint undercutting. Filiform Beach, N.C., 80 ft (24 m) from the ocean [Coburn 1978]. The corrosion occurred in several cases, and there was practically first test involved a 15-year exposure of 10 paint systems applied no difference between carbon and weathering steels. These re- on carbon and weathering steels. The specimens measuring sults were said to be in line with those obtained by other insti- 8 in. x 12 in. (0.20 m X 0.30 m) were sandblasted to white tutes in the Netherlands. metal and painted with a 4.5 to 6.5 mils (115 to 165 j.tm) thick two-coat or three-coat system of alkyds, vinyls, vinyl-acrylics, Van Eijnsbergen concluded that container roofs and other chlorinated rubber or catalyzed epoxy and polyurethane com- parts made of weathering steel must receive the same coating positions. The paint began to deteriorate at the cut edges, and as other types of steels; that is gritblasting, followed by a. coat the rust gradually undercut the coating on the flat faces. Table of, preferably, zinc-rich primer and one or two coats of an epoxy 51 shows that all paint systems were more durable on the weath- ester or chlorinated rubber-based top coat with a total thickness ering than on the carbon steel. of 3 to 4 mils (75 to 100 .tm). Inorganic zinc-rich coatings have The second test performed in Kure Beach, N.C., determined a unique ability to prevent spreading and creeping of rust from whether a surface free of mill scale is required for optimum bared areas adjoining the paint coat. It must be appreciated that paint performance. Specimens measuring 12 ft x 1 ft (3.66 m it takes time for the weathering steel oxide films to stabilize. x 0.30 m) were fabricated from a carbon steel and a weathering Three years of exposure is just about the point where further steel used for sheet piling in seawater containing 0.5 percent attack is stifled. copper, 0.5 percent nickel, and 0.1 percent phosphorus. Matched Scantlebury tested the corrosion inhibiting properties of pairs were sandblasted and hand wire brushed, coated with a aqueous extracts of zinc phosphate pigment on carbon steel and variety of thick mastics and thin coating systems and X-marked a low-alloy copper-bearing steel [Scantlebury 1980]. He found at the lower 5 ft (1.52 m). The plate specimens were mounted that durability of the paint was identical for both steels. on a rack facing south at a 30-deg angle and exposed 80 ft (24 Tinklenberg [1982] initiated a study to learn whether coating m) from the ocean in Kure Beach, N.C. The test showed that systems applied to nonweathered or fresh A36 and A588 steel removing the mill scale by sandblasting ensured good paint panels performed differently from each other in a chloride en- performance, regardless of substrate steel composition. It also vironment. The fresh panels were blast-cleaned to a near-white showed that paints were more durable on hand wire-brushed (SSPC 10) condition, coated with the eight paint systems listed weathering steel than on similarly cleaned carbon steel. in Table 52, and placed in a salt-fog cabinet for the durations indicated in the table. He found no significant difference between One Japanese, two European, and one American study re- the performance of the various coating systems on shop-coated ported that paint systems applied on carbon and weathering A36 and A588 panels in a moist chloride environment (salt fog). steels were equally durable [Nakayama 1972, Van Eijnsbergen According to the ratings in Table 52, the zinc-rich systems 1979, Scantlebury 1980 and Tinklenberg 1982]. performed best, followed by the urethane, alkyd and mastic Nakayama showed that after one-year exposure the damaged systems. parts of painted atmospheric corrosion resistant steel corroded On the basis of the published data, there appears to be con- less than those of carbon steel [Nakayama 1972]. However, after vincing evidence that when a variety of paint systems are applied 4-years exposure, no distinct difference in paint performance was observed. According to Nakayama, the use of atmospheric corrosion resistant steel does not necessarily simplify painting procedures. Van Eijnsbergen, whose interests related to galvanized steel Table 52. Paint systems applied to freshly blast-cleaned A36 and A588 and shipping containers, investigated paint performance on both steels and tested in salt-fog cabinet. [Tinklenberg 1982] weathering steel and carbon steel, especially when damaged locally by scratches and dents [Van Eijnsbergen 1979]. Salt Fog, A 36 vs. A588 No. Generic Type hours Performance He also investigated the claim that the use of an anticorrosive primer is unnecessary when applying a paint system on weath- 1 Inoranlc zinc-rich, vinyl 4,000 Equal-very good ering steel surfaces, and that only a top coat is required. For 2 Organic zinc-rich, epoxy 4,000 Equal-excellent this purpose, he exposed carbon and weathering steel specimens oora to the marine-industrial atmosphere of The Hague, Netherlands. 3 Aimninun epoxy mastic 2,500 A588 worse-both p All specimens were gritblasted and painted with one or two 1.2- 4 4-coat no lead alkyd 2,500 A36 worse-both poorb to 1.6-mils (30- to 40-sm) thick epoxy top coats. The results 5 3-coat urethane, epoxy 2,500 Equal-both good after 3 years of weathering were: 6 4-coat urethane, epoxy 2,500 Equal-both good

7 3-coat red lead al kd 2,500 Equal-both poor to fair On all one-coat systems, considerable rust formed, irre- 8 4-coat no lead alkc 2,500 Equal-both fair spective of type of steel. On all two-coat systems, slight or no rust formed. Notes:

On the back side of the specimens, the same degree of rust aihe A588 panel had a thinner dry film thickness which caused the occurred. But on the areas where the specimens were supported poorer rating. by wooden beams rusting was more pronounced, especially with bThe A36 panel developed a large crack at 2,000 hours which lowered its the one-coat systems. rating. In the 30- to 40-mils (0.8 to 1.0-mm) wide scratches pen- Conditions (a) and (b) are not related to the type of steel. etrating to the steel, there was either no difference or only a COifferent pigment system than in No. 4. 128 on both carbon steel and weathering type steel compositions, m) thick Prepalene, is a highly permeable film of acrylic resins and exposed in different environments long enough to exhibit that prevents dislodging of the initial rust products while the degradation, the carbon steel panels will be the first to degrade. oxidation reactions progress. The previously cited 15-year exposure tests, performed by Co- The primary coat is gradually absorbed by the oxide, and the burn 80 ft (24 m) from the ocean in Kure Beach, N.C., confirm secondary coat vanishes with time. the results from the tests by Copson and Larrabee, in which As an example of how the Weather-Coat system is applied the longer life of paints on corrosion resistant steels was first in practice, the treatment of the three-span continuous box girder demonstrated [Coburn 1978, Copson 1959]. of the Nadahama Bridge, Hyogo Prefecture, Japan, is described Some investigators tested paint systems over shorter periods below. The following steps were carried out in the fabrication of time and were unable to reach the same conclusions, because shop: their exposure periods were too short for the paint to degrade Blast cleaned steel surfaces. on the carbon steel panel. Other investigators accelerated the Within 6 hours after blast cleaning, sprayed at least three testing time by imposing on the paint systems artificial envi- times Weather-Coat #1000 solution at 122 F (50 C). Allowed ronments in which the weathering steel was incapable of de- 15-min drying time between two coats. veloping a protective oxide coating-the reason being that the Rinsed surfaces with clean water immediately after test cabinets did not permit the necessary alternate wetting and step 2. drying cycles in the presence of sulfur oxides. Such artificial Allowed surfaces to dry naturally. environments can only discriminate between paint systems, re- Masked surfaces that were to be treated with Prepalene gardless of the substrate. (friction-type joints, etc.). Weathering steels enhance the service life of paint, because After 30 min to 24 hours drying, sprayed a 1.2-mils (30- the protective oxide coating in scratches and thin spots is suf- p.m) thick film of dark brown Prepalene PS. Allowed it to dry ficiently thin to avoid undercutting the adjacent intact paint for 30 mm. film. With carbon steel, the voluminous rust continuously de- Sprayed a 0.8-mils (20-)xm) thick film of red brown Pre- stroys the adhesive bond between the paint and the surface of palene R. the steel. On the other hand, when environmental conditions Air dried surfaces for more than 12 hours. are so bad that the weathering steel would have to be painted, Inspected appearance and average dry film thickness (min- A572 steel would be a preferred economic choice. imum 2 mils (50 p.m). Should a decision be made to paint the weathering steel, the same surface preparation procedures and standards applied to After the steel was erected, the following additional steps were carbon steel should also be applied to the weathering steel. The carried out at the job site: service life of the paint is directly related to the amount of surface preparation. Removed surface contamination by concrete, oil, and dust. Removed rust from bolt-heads, etc. RUST STABILIZING TREATMENT Repaired damaged areas by repeating steps 2 to 9. Sprayed a 0.8-mils (20-p.m) thick film of red brown Pre- Most weathering steel bridges in Japan were either painted palene R finish. or treated with an oxide stabilizing coating. The latter is designed Air dried surfaces for more than 12 hours. to: (1) accelerate the formation of a dense and stable oxide Inspected appearance and average dry film thickness coating; and (2) prevent dislodging and outflow of the initial (minimum 2.8 mils (70 gm)). rust. Such treatments are being used to ensure that a protective oxide coating develops in the Japanese climates often charac- "Weather-Coating" is a lengthy and costly process. It adds terized by high humidity and high temperature, and that the about 10 to 20 percent to the basic cost of a weathering steel concrete piers and abutments are not stained by rust-bearing bridge in Japan [Yamada 1983a]. The Shiribetsu Railway Over- water running off steel members. pass Bridge, the first to be built of Weather-Coated weathering To counter the aforementioned problems, Nippon Steel Corp. steel, was completed in November 1968. The exposure time of of Japan developed the Oxicoat and Weather-Coat surface treat- such bridges has not been long enough to conclusively evaluate ment systems [Nippon 1974, Nippon 1976]. The Oxicoat system the performance of the Weather-Coat process [Yamada 1983a]. mainly accelerates the formation of a stable oxide coating, Questions remain about how bridges treated with Weather-Coat whereas the Weather-Coat system mainly inhibits outflow of perform in environments in which they are subjected to salt rust. Both systems are sacrificial. They are only applied once, spray. The Nakashio Bridge in Tokyo, Japan, is an example of before the steel structure is erected. a Weather-Coated bridge constructed over the sea. In this case, The Weather Coat system consists of two coats. The primary after 2 years of exposure, a stable oxide coating was forming, coat is 0.4 to 0.6 mils (10 to 15 m) thick. Its main component except on internal beams which are rusting faster than antici- is crystal zinc phosphate, Zn3 (PO4)3. H2O. Its function is to pated [Nippon Steel]. accelerate the formation of the stable oxide coating. The primary coat is applied either by immersion or by spraying. The immersion method is suited for shop application to sheets and small structural members at relatively high temperatures. The REMEDIAL PAINTING OF EXISTING BRIDGES spray method is suited for large structural members and for field touch-up of damaged coats. The Michigan Department of Transportation and the Steel The secondary coat, consisting of 0.16 to 1.6 mils (4 to 40 Structures Painting Council are currently working on conduct- 129 ing test programs on cleaning and painting corroded weathering Table 53. Steel panels used to evaluate applicability of salt-fog cabinet steels. The only results available to date are those from the testing to service conditions. [Tinklenberg 1982] Michigan study [Tinklenberg 1982]. In comparison with shop coating new steels, the problem with Type of Steel Exposure Condition remedial painting of severely corroded and pitted weathering A588 Environmentally exposed (Eight-Mile Road Bridge) steels lies in effectively removing all rust and salt trapped in A588 Exposed uncoated in salt-fog cabinet for 1000 dimples and pits. Unless this is achieved, paint life is reduced. hours Continuing the work cited earlier in this chapter under "Initial A36 and A588 Exposed outdoors in rural atmosphere Painting of New Bridges," Tinklenberg conducted three tests on remedial painting. In the first, A36 and A588 steel panels A588 Not exposed were exposed as described in Table 53. The environmentally exposed panels were cut from link plates that were removed from the Eight-Mile Road Bridge over 1-75 in Detroit, Michigan. Table 54. Evaluation of coating systems on weathered A36 and A588 The panel material had been exposed in service for about 12 steels in salt-fog cabinet. [Tinklenberg 1982] years. The joint in this area leaked badly and the steel was heavily corroded and pitted. The weathered panels were sand- No. of Systems Type of Coating blasted to a near-white (SSPC-10) condition and painted with the nine coating systems given in Table 54. Thereafter, they 1 Four-coat lead free alkyd were exposed in a salt fog cabinet for 4,000 hours in accordance 2 Single-component organic zinc-rich primer with epoxy top Coat with the requirements of ASTM Specification B-1 17. Tinklen- 3 Multi-component (usually three) epoxy zinc-rich berg concluded the following: primer with epoxy topcoat

1 One-coat epoxy mastic If a problem is encountered with a laboratory panel, it 1 Single-component inorganic zinc-rich primer with could be worse with environmentally exposed A588 steel. vinyl topcoat

Three of the nine coating systems blistered uniformly on 1 Two-component inorganic zinc-rich primer with all five panels. Three of the coating systems blistered only on epoxy topcoat the face that had been environmentally exposed and retained salt residue in the surface crevices. Five of the zinc-rich compositions performed very well. Table 55. Surface preparation of environmentally exposed A588 panels They showed little if any corrosion damage in the 4,000-hour removed from Eight Mile Road Bridge over 1-75 in Detroit, Michigan, and coated with systems given in Table 56. [Tinklenberg 1982] test. Natural weathering of A588 steel cannot be reliably du- No. of Panels Surface Preparation plicated in the laboratory with relatively short-time salt fog exposure. 1 Blast-cleaned to near-white condition Coating systems on environmentally exposed A588 panels 1 Brush-off blasted, power washed and reblasted to tend to fail more rapidly than those on unexposed panels. near-white condition

With proper preparation and application, it is possible to 1 Near-white blasted, power washed, and reblasted get excellent corrosion protection of A588 steel, even if it has to near-white condition been exposed to a leaky bridge joint environment. 1 Near-white blasted and exposed for 10 minutes to 100% relative humidity to produce the "green On the basis of the observations from this group of tests mold phenomenon" and the tests described under "Initial Painting of New Bridges," 1 Gradient blasted from brush-off to near-white one can conclude that the paint life is likely to be shorter on condition weathering steel previously exposed to even ideal environmental conditions, than on new steel. Table 56. Coating systems applied to steel panels given in Table 55 and to near-white-blasted A36 panels. [Tinklenberg 19821 In the second group of tests on remedial painting; Tinklenberg examined the effects of different surface preparation methods. No. of Systems An additional objective was to begin performance evaluation of Tested Type of Coating System Rating various generic types of coatings. The nonweathered control 2 Multi-component organic zinc-rich with 1 panels, made of A36 steel, were blast cleaned to a near white epoxy intermediate coat and an aliphatic (SSPC-10) condition. The weathered A588 steel panels were cut urethane topcoat from hanger plates that had been removed from the Eight-Mile 1 Single component organic zinc-rich with 2 Road Bridge. The latter were prepared for painting by one of a polyainide epoxy topcoat the five methods described in Table All panels were then 3 SIngle component inorganic zinc-rich 3 55. primer with polyamide-epoxy topcoats painted with the 10 coating systems given in Table 56 and placed 1 Moisture-cured urethane with a polyamide 4 in the salt-fog cabinet. Based on a preliminary analysis of the epoxy topcoat data, Tinklenberg concluded the following: 1 Polyamide epoxy primer with aliphatic 5 urethane topcoat

1. The washing cycle between the two blasting operations 4-coat chlorinated rubber system 6 had little, if any, effect on paint performance. The green mold 1 4-coat no lead no chromate alkyd system 7 that formed on A588 steel had little effect on performance. 130

Blistering was generally worse in brush-off and commer- qualified products list and specify cleaning procedures for salt- cially blasted areas. contaminated weathering steels. To date, it would appear that In general, the appearance was worse on the side of the one-coat systems, alkyd systems, and chlorinated rubber systems hanger plate that had been exposed to the environment. This is are not suitable. Organic zinc-rich systems appear to protect of concern as the interior surface was more severely pitted and best. They seem to tolerate a wider variation in surface prep- much worse in initial appearance. Also, the outside surface more aration. Some of the other systems may be satisfactory, especially closely resembled the exposure and appearance of the entire when cost is considered. It would appear that remedially painted bridge surface. bridges can perform well when suitable cleaning procedures and After 3,000 hours of salt-fog cabinet exposure, the generic coating systems are selected [Tinklenberg 1982]. coating systems that were applied to the A588 panels ranked Table 58 describes the surface preparations and coating sys- in the order given in Table 56. The systems ranked 3rd, 4th, tems that were used for the remedial painting of nine weathering and 5th performed about equally. steel bridges. The bridge on Tuscarawas County Road 99 over Sugar Creek was painted 2 months after construction, as a A conclusive performance evaluation must await the final anal- precautionary measure, when it became apparent that the ysis of cost, film thickness, and ease of application under field other weathering steel bridge in the County, on Road 37 over conditions. Little Still Water Creek, was corroding severely and had to Based on the results of the two previously described test series, be rehabilitated. Tmklenberg initiated an extensive qualification program for Other severely corroded weathering steel bridges are sched- coating systems. As before, the unexposed control panels of A36 uled for remedial painting, namely: four bridges on Big Salt steel were blasted to a near-white (SSPC-lO) condition. The Lake(Road, Prince of Wales Island, Alaska; and several bridges weathered A588 panels were cut from hanger plates removed in Michigan. In these cases, the selection of painting systems is from a badly corroded joint area of the Grand Boulevard Bridge awaiting the results of the on-going Michigan study. over the C&O Railroad in Detroit, Michigan. They were cleaned Initial experience with remedial painting shows that it takes to a near-white, white or gradient blasted condition. All steel more than the usual amount of sand to blast weathering steel panels were then coated with the 22 systems given in Table 57. clean, and more than the usual amount of paint to coat it The tests are in progress. When all results become available, [Baron 1983]. the Michigan Department of Transportation will generate a

Table 57. Description of the 22 coating systems currently being eval- Table 58. Systems used for remedial painting of weathering steel bridges. uated in the salt-fog cabinet to determine their relative performance when applied to near-white, white, and gradient-blasted A588 steel panels and near-white blasted A36 steel panels. [Tinldenberg 1982] Year Opened Year to traffic Painted Surface Preparation and Paint System No. of Systems Type of Coating System

3 Single-component inorganic zinc-rich primers with Brand Road Bridge, Franklin County, Ohio, [Circle 19831: hi-build vinyl topcoats 1979 1983 - Sand blasted to bare metal 2 Single-component inorganic zinc-rich primers with - Two Coats of red-lead primer polyamide epoxy topcoats - One coat of oil-alkyd paint

Single-component inorganic zinc-rich primer with a Bridge on County Road 37 over Little Still Water Creek, Tuscarawas polyamide-epoxy intermediate coat and an aliphatic County. Ohio, [Smith 1983]: urethane topcoat 1973 1979 - Sand blasted 1 Multi-component inorganic zinc-rich primer with hi- - One coat of zinc-rich primer build urethane topcoat - One coat of modified alkyd Multi-component inorganic zinc-rich primer with a hi- resin paint build epoxy intermediate and an aliphatic urethane topcoat Bridge on County Road 99 over Sugar Creek, Tuscarawas County. Ohio, [Smith 1983]: 1 Single component organic zinc-rich primer with a polyamide epoxy topcoat 1979 1979 - Wire brushed as needed - One coat of zinc-rich primer 3 Multi-component organic zinc-rich primers (one of which is a urethane) with polymnide epoxy topcoats - One coat of rubberized paint

1 Multi-component organic zinc-rich primer with a polyamide epoxy intermediate and aliphatic urethane 1983 - Wire brashed top coat topcoat - Touched up primer, as needed 2 Moisture-cured urethanes with an aliphatic urethane - One coat of industrial safety topcoat green paint

1 Moisture-cured urethane with a polyamide epoxy Doullut Canal Bridge, Empire, Louisiana [Garrido 1983, Dunn 1983]: intermediate and an aliphatic urethane topcoat

Polyamide epoxy primer with an aliphatic urethane 1 1975 1983 - Sand blasted to near-white topcoat condition (SSPC-10) Synthetic resin primer with a polyamide epoxy topcoat 1 - Two coats of organic zinc-rich 1 Aluminum-filled epoxy-urethanemastic primer, 115 cm (4.5 oils) thick - One coat of vinyl, 38 on 1 Aluminum-filled epoxy mastic (1.5 mils) thick 1 Zinc-filled and aluminum-filled epoxy mastic with hi-build aliphatic urethane topcoat Six Bridges in Butler County, Ohio [Brooks 19831: After 1971 1983 - Sand blasted 1 4-coat, no lead, no chromate alkyd - One coat of asphalt-base paint 131

CHAPTER TEN

CONNECTIONS

ARC WELDING Table 59. Weldabiity rating based on carbon equivalent values.

Most welded connections of bridge members are made by the Weldability Range of Carbon Equivalent Values shielded-metal-arc (SMA), submerged-arc (SAW) and flux- Rating Producer A Producer B cored arc (FCA) processes. A few bridges have electroslag Excellent CEV < 0.40 CEV < 0.35 welded flanges, but the use of this process is no longer permitted Very good 0.36 < CEV < 0.40 Good 0.41 < CEV <0.45 0.41 < CEV < 0.45 for main structural tension members. Fair 0.46 < CEV <0.52 0.46 < CEV <0.50 The weldability of steel is a complex property involving: (1) Poor CEV > 0.52 CEV > 0.50 the sensitivity to weld and heat-affected-zone (HAZ) cracking; and (2) the toughness in both the weld and HAZ, as required by service conditions and temperatures. The first is a welding- values are much lower because the content of all alloying ele- fabrication problem, the second a service problem. Most welding ments need not be simultaneously maximized to achieve the cracks are termed cold cracks or delayed cracks because they desired strength and corrosion properties. The carbon equivalent usually occur at temperatures lower than 392 F (200 C). They of a specific heat of weathering steel can be determined from consist of toe cracks, root cracks, and underbead cracks, and the chemical composition given in the mill report. are caused by a combination of: (1) lower ductility due to hard- Hydrogen embrittlement of the HAZ can be minimized by ening of the HAZ; (2) embrittlement due to diffusion of hy- controlling the hydrogen content of the weld metal. This is drogen into the HAZ, and (3) residual welding stresses in high achieved by using low-hydrogen electrodes, proper drying of restraint details and thick plates. electrodes fluxes and gases, and other suitable techniques for The so-called carbon equivalent (CE) is a convenient index of the cold cracking susceptibility of welds. Experimentally de- minimizing the hydrogen content of the molten weld metal. termined CE formulas take several forms, depending mainly on Residual welding stresses and applied stresses, in combination their applicability to certain types of steel. The Dearden-O'Neill with cold cracks, can produce a condition critical for unstable formula, developed for the safe welding of carbon-manganese crack extension, particularly in thick sections nd restrained steel, shows how the various alloying elements influence hard- welds. To control this possibility, A242, A588, and A709 Grade enability. 100W steel plates are preheated before welding, so as to retard the rate of cooling after welding [AWS 1984]. Thicker sections dissipate heat faster and are, therefore, preheated to higher Carbon Equivalent (CE) temperatures. The higher the hardenability of the steel, the Mn Cr + Mo + V + Ni + Cu (56) =C+ — — + greater is the need for preheating, because higher hardenability 6 16 promotes the formation of martensite and increases the probability of cold cracking. Heat input is also important because The numbers to be substituted for the elements are the alloy it affects the rate at which the weld cools after the arc has passed contents in weight percent. As Eq. 56 shows, some elements on. The more heat is introduced by the arc, the slower the used for enhancing atmospheric corrosion resistance and for cooling rate. Heat input is related to arc voltage, welding current, increasing strength through solid-solution or precipitation and travel speed. strengthening also increase the suceptibility to cold cracking. Another form of cracking, called solidification or hot crack- In HSLA steels, this adverse effect is offset in part by limiting ing, may occur when during the solidification of the weld metal the carbon content. the interdendritic liquid (the last region to freeze) freezes at a Table 59 shows weldabiity rating scales in terms of CE, used substantially lower temperature than the bulk solidified dendrite. by two Canadian steel producers [SCC 1973, DFSL 1975]. They Under these conditions, shrinkage stresses produce interden- are quite similar. The first producer reported 0.44 mean CE dritic cracks. This type of cracking is most commonly caused and 0.50 maximum CE for its multiple-alloy weathering steel. by the presence of low-melting iron sulfides and alloy sulfides According to Table 59, the mean CE value implies good weld- that wet the dendritic surfaces. Carbon, phosphorus, and sulfur ability. are harmful alloys. The A588 specification, for example, limits The Canadian Standard G40.21 for Structural Quality Steel the maximum content to about 0.20 percent C, 0.04 percent P cites Eq. 56. The ASTM specifications A242, A588, and A709 and 0.05 percent S. Hot cracking is controlled by limiting the and the product literature issued by American steel manufac- amount and type of sulfides that form and the amount of minor turers do not mention CE values. If one were to substitute in alloying elements that may promote it. Manganese-to-sulfur ra- Eq. 56 the maximum alloy contents for the various grades of tios of 30 or more help to prevent hot cracking of weld metal. weathering steel, as specified in the A588 specification, one Maximum amounts of manganese and sulfur, listed in the A588 would obtain CE values no greater than 0.71 percent. Actual specification, give about this ratio. 132

The previous discussion briefly highlighted some of the more temperature is below the temperature given in this table for the important factors that need to be considered in welding low- thickness of material being welded, it shall be preheated (except alloy steels. They help to emphasize the importance of following as otherwise provided) in such manner that the parts on which proper welding procedures so that a sound weld is obtained. the weld metal is being deposited are above the specified min- Weathering steels supplied to ASTM Specifications A242, imum temperature for a distance equal to the thickness of the A588, and A709 Grades 50W and 100W are weldable. Their part being welded but not less than 3 in. (76 mm) in all directions weldability is comparable to that of their A572 and A5 14 coun- from the welding point. Quenched and tempered steel A709 terparts. Weathering steels can be welded to other structural Grade 100 W steels should not be preheated or welded at an steels without difficulty. interpass temperature above a maximum temperature, depend- The welding technique is of fundamental importance. The ing on grade and thickness, or the steel would lose some of its adopted procedure must be suitable for the steel being welded strength and toughness. The AASHTO specification requires and its intended use. higher preheat temperatures for fracture critical nonredundant members made of A588 and A709 Grade 100W steels than for SPECIFICATIONS redundant members made of the same steels [AASHTO 1978].

Weathering steels, especially those for bridge applications, are ELECTROSLAG WELDING weldable with the use of good welding practice and the proper choice of filler metal. Carbon steel or low-alloy steel electrodes The electroslag welding (ESW) is a butt-welding process or electrode-flux combinations may be used, provided that the wherein coalescence is produced by molten slag which simul- welding procedure ensures a sound weldment that matches the taneously melts the filler and the parent metals. The slag shields mechanical properties and the atmospheric corrosion resistance the full cross-section of the joints and remains molten by its of the base steel. resistance to electric current passing between the electrode and All weathering steels require a minimum preheat and inter- the metal. Welding is done in the vertical or near vertical po- pass temperature. This is a characteristic of all high strength sition, and joints are completed in a single pass for any thickness steels in thick sections. In addition, A709 Grade 100W steel by using one or more electrodes. has limitations on maximum preheat and interpass temperatures As the molten slag pool and the weld pool rise in the joint, and limitations on welding heat input. The provisions of the they are contained by two molds or shoes made of copper or American Welding Society (AWS) "Structural Welding Code" sometimes steel plates. The copper shoes are either solid (dry (latest edition) and the AASHTO "Standard Specifications for shoe) or hollow for cooling water circulation (cooled shoe). One Welding Structural Steel Highway Bridges" (latest edition), in study indicates that reducing the weld cooling rate during and conjunction with the steel producer's recommendations, ensure after welding by the use of dry versus cooled shoes reduced the that the filler metal, preheat temperature, and other relevant number of grain boundary separations in an A588 electroslag welding parameters provide suitable welded joints for the in- weldment [Konkol 1979]. Other reports indicate that slow cool- tended application [AWS 1984, AASHTO 1981]. ing of electroslag weldments prolongs the thermal cycle and results in "coarse casting" of the weld metal and the HAZ. Filler Metal Requirements Electroslag weldments can be made with consumable and nonconsumable guide tubes. The consumable guide tube method is The filler metal requirements for welding of weathering steels used more often because of equipment and set-up simplicity. are given in Table 60. Both AWS and AASHTO specify low The main advantage of electroslag welding is the savings in hydrogen electrodes for SMA welding of weathering steels. Cov- manpower, time, and welding consumables. In addition, the joint ered carbon steel electrodes (AWS AS. 1) and bare carbon steel does not require machine edge preparation, and a proper welding electrode-flux combinations (AWS AS. 17) listed in Table 60 are procedure eliminates the necessity for preheating. The very high used for(l) underlying passes of multiple-pass weldments which heat input and prolonged thermal cycle, with accompanying need not be corrosion resistant; (2) single-pass SMAW up to slow solidification and cooling rates, results in a "coarse casting" 0.25-in. (6.4-mm) maximum weld dimension; (3) single-pass type of weldment structure with anisotropic and nonhomoge- SAW, GMAW, and FCAW up to 5/ 6-in. (8.0-mm) maximum neous large-grained weld metal and extremely large HAZ. Sev- weld dimension; and (4) painted structures. Covered low-alloy eral studies showed that the HAZ had at the center both a lower steel electrodes (AWS A5.5) and bare low-alloy steel electrode- yield point and a lower Charpy V-notch energy than the sur- flux combinations (AWS A5.23) listed in Table 60 are used for rounding weld metal, and sometimes lower than the base metal exposed, bare applications of ASTM A242, A588, and A709 [Paton 1962a, 1962b, Culp 1976; Noel 1972]. Others have con- Grade SOW steel requiring weld metal with atmospheric cor- firmed these results and have further reported a loss in ductility rosion resistance and coloring characteristics similar to the base in the weld metal as compared to the base metal [Norcross 1965, metal. At least two layers of the exposed surfaces and edges of Woodley 1966, Campbell 1970]. multiple-pass weldments must be deposited with these electrodes Following the failure near Pittsburgh, Pa., of the Interstate provided the underlying layers are deposited with one of the 1-79 bridge made of A588 steel, the Federal Highway Admin- filler metals specified in Table 60. istration banned the use of the electroslag welding process for main structural tension members on any Federal-aid project. Minimum Preheat Interpass Temperature The AWS "Structural Welding Code-Steel" also prohibits the use of this process for welding quenched and tempered steels The minimum preheat and interpass temperatures for welding and for welding bridge members subjected to tension or stress of weathering steels are listed in Table 61. When the base metal reversal [AWS 1984]. 133

Table 60. Filler metal requirements for welding weathering steels. [AWS 1984]

ASTM Steel Matching Filler Filler Metal Requirements for Designations a,b Metal Requirements Exposed Base ApplicationsC Shielded Metal Submerged Shielded Metal Submerged Arc Arc Arc Arc A242d, A588d and AWS A5.1 AWS A5.17 or A5.23 AWS A5.5 AWS A5.23' A709 Grade 50W E7015 F7X-EXXX E7018-W F7AX-EXXX-W E7016 E8018-W E7018 E8016-C3 or E8018-C3 F7AX-EXXX-Ni 1e E7028 E8016-Cl or E8010-Cl F7AX- EXXX-Ni 4e E8016-C2 or E8016-C2 E7016-CIL or E8018-CIL F7AX-EXXX-Ni 2e E7016-C2L or E8018-C2L F7AX-EXXX-Ni 3e E8018-B2L

A709 Grade 100W AWS A5.5 AWS A5.23 t<63 mm E11015 F11X-EXXX1' (t<2.5 in) E11016 E11018

A709 Grade 100W AWS A5.5 AWS A5.23 63 mm < t < 102 mm E10015 F10X-EXXX (2.5 in < t < 4 in) E10016 E10018

Notes

In joints involving base metals of two different groups low-hydrogen filler metal requirements applicable to the lower strength group may be used. The low-hydrogen processes used shall be subject to the technique requirements applicable to the higher strength group.

When welds are to be stress-relieved, the deposited weld metal shall not exceed 0.05 % vanadiimi.

Deposited weld metal shall have the following chemical composition: C, max 0.12%, Mn 0.50%/1.30%, P, max 0.04%, S max 0.04%, Si 0.35/0.80%, Cu 0.30/0.75%, Ni 0.40/0.80%, Cr 0.45/0.70%.

Special welding materials and procedures (e.g., E80XX low alloy electrodes) may be required to match the notch toughness of base metal (for applications involving impact loading o low temperature), or for atmospheric corrosion and weathering characteristics.

The use of the saie type of filler metal having next higher tensile strength as listed in AWS specifications is permitted.

Deposited weld metal shall have a minimum impact strength of 27.1 J (20 ft. lb) at -18'C (O'F) when Charpy V-notch specimens are used. This requirement is applicable only to bridges.

See AWS specifications for filler metal requirements for gas metal arc and flux cored arc welding.

134

Table 61. Minimum preheat and interpass temperature for welding of weathering steels. [AWS 1984]

Thickness of ASTM Steel Welding Thickest Part Minimum Designation Process at Point of Welding TemperaturedC m(in) 'C('F)

A242, A588 and SMAW with low t < 19 (0.75) None A709 Grade 50W hydrogen electrodes, 19 (0.75) < t < 38 (1.50) 10 ( 50) SAW, GMAW and FCAW 38 (1.50) < t < 64 (2.50) 66 (150) 64 (2.50) < t 107 (225)

A709 Grade 100Wb SMAW with low t < 19 (0.75) 10 ( 50) hydrogen electrodes, 19 (0.75) < t < 38 (1.50) 50 (125) SAW with carbon or 38 (1.50) < t < 64 (2.50) 80 (175) alloy steel wire, 64 (2.50) < t 107 (225) neutral flux, GMAW or FCAW

Welding shall not be done when the ambient temperature is lower than -18'C (O'F). When the base metal is below the temperature listed for the welding process being used and the thickness of material being welded, it shall be preheated (except as otherwise provided) in such manner that the surfaces of the parts on which weld metal is being deposited are at or above the specified minimum temperature for a distance equal to the thickness of the part being welded, but not less than 76 ran (3 in) in all directions from the point of welding. Preheat and interpass temperatures must be sufficient to prevent crack formation. Temperature above the minimum shovi may be required for highly restrained welds.

The maximign preheat and interpass temperature shall not exceed 205'C (400F) for thickness up to 38 ran (1.5 in) inclusive, and 230'C (450'F) for greater thickness. The heat input shall not exceed the steel producers recommendations.

In joints involving combinations of base metals, preheat shall be as specified for the higher strength steel being welded.

The bar on the use of electroslag welding in bridges is basically Table 62. Tensile requirements for high strength bolts. a technological matter. It was believed that the electroslag welding technology had not reached the point where a defecC free Proof Stress, Alternative Length Proof Stress, welding could be made. Separations larger in size than the Tensile Measurement Yield Strength critical flaw were detected. Reinspection of electroslag welds in ASTM (A.ASRTO) Bolt size Strength Method Method Specification m (in.) MPa (ksi) MPa (ksi) MPa (ksi) highway bridges revealed a high percentage of welds containing

unacceptable weld defects, and some were judged to be struc- P325 12 < t < 25 825 585 635 turally unsafe [Atteridge 1982]. (M164) (1/2 < T < 1) (120) (85) (92) In order to yield consistent and defect-free welds, substantial 29 < t < 38 725 510 560 changes are needed in ESW fabrication and inspection tech- (1-1/8 7 t (105) (74) (81) nology. One study sponsored by the FHWA developed and ! 1-1/2) assessed fabrication methods aimed at improving ESW tech- A490 12 < t < 38 1035-1170 825 895 nology [Atteridge 1982]. The results indicated that defect-free (M253) (1/2 < t (150-170) (120) (130) ESW could repeatedly be made if certain welding procedures 1-1/2) were followed. It was concluded that commercial operations tend to weld at too low a slag level and at too high a current. The customary Use narrow gap (3/4 in. (19 mm)) high current (1,000 Amp) technique of adding relatively large amounts of flux to the slag welding in combination with a winged guide tube to: (a) decrease pool accentuates the welding problem. Relatively simple solu- the specific heat input; (b) increase the welding speed; and (c) tions for quality control were suggested by Atteridge et al. as most importantly, to improve the mechanical properties of the follows [Atteridge 1982]: "weld centerline at midthickness location. Use a thin quartz shroud for guide tube shielding and Use a sliding shoe(s), as in the nonconsumable welding intense mechanical stirring of the weldpool to achieve grain method, to allow constant access to the slag pool. refinement and homogeneous weldment throughout the length. Add a simple but accurate current monitoring device. This however requires further research. Continuously add flux.

Also suggested were the following fabrication changes to up- HIGH-STRENGTH BOLTED JOINTS grade the technology and reduce anisotropy and nonhomo- geneity of electroslag weidments: High-strength bolts are furnished to the requirements of ASTM specifications A325 and A490. Both are made of 1. Use a consumable winged guide tube instead of a cylin. quenched and tempered steel. Their minimum tensile require- drical guide tube to eliminate guide tube misalignment. ments are summarized in Table 62. 135

Table 63. Chemical requirements for Type 3 bolts based on product analysis.

Class Composition %

C Mn P, max S, max Si Ni Cr Mo, max Cu V, mm

A325 Type 3 Bolts (M164

A .31 .86 .045 .055 .13 .22 .42 .22 .42 1.24 .32 .48 .68 .48

B .36 .67 .06 .25 .47 .47 .07 .17 .50 .93 .125 .055 .55 .83 .83 .43

C .14 .76 .040 .045 .13 .22 .27 .17 010 .26 1.39 .32 .53 .53 .53

0 .14 .36 .045 .055 .20 .47 .45 .11 .27 .26 1.24 .55 .83 1.05 .53

E .18 .56 .045 .045 .13 .27 .55 .27 .27 1.04 .32 .63 .95 .63

F .19 .86 .045 .045 .13 .17 .42 .17 .26 1.24 .32 .43 .68 .43

A490 Type 3 Bolts (M253)

.19 .37 .045 .055 .17 .42 (a) .63 .55 min max max min mm. max

Note:

a. 0.17% Ni min can be replaced by 0.14% Mo mm.

The bolts come in three tyyes. Type 1 and Type 2 are intended weathering steel has indicated that if the stiffness of the joint for use in painted construction. Type 3 bolts have atmospheric is adequate and the joint is tight, the crevice between two contact corrosion resistance and weathering characteristics comparable surfaces seals itself as corrosion products form around the pe- to those of weathering steels covered in the A242, A588, and riphery of the joint. However, if the joint design does not provide A709 specifications. The chemical requirements are given in sufficient stiffness, continuing crevice corrosion and subsequent Table 63. A325 Type 3 bolts come in five classes, each with a accumulation of corrosion products in the crevices induce ex- distinct chemical composition. A490 Type 3 bolts come in one pansion forces which can: (1) deform the connected elements, general class, with minimum and maximum limits set on each and (2) cause large tensile loads on the bolts. alloying element. Brockenbrough inspected weathering steel transmission tow- High-strength bolts must be tightened to a minimum tension ers with joints connected by bolts furnished to another speci- equal to 70 percent of the specified bolt tensile strength. The fication, but having a measured tensile strength approaching high installation forces introduce high prestresses between the that of A325 bolts [Brockenbrough 1983]. The bolts were only contact surfaces in a region equal to about two bolt diameters. snug tightened. It was found from these inspections that the Within this region, the joints are well seated. The joints are plates between two bolts did not significantly bow when the designed to resist shearing forces either as friction-type or bear- pitch did not exceed 14 times the plate thickness or 7 in. (178 ing-type connections. Slip along the contact surfaces is prevented mm). Similarly, the plate edges did not significantly lift up when in the former but not in the latter case. the edge distance was limited to 8 times the plate thickness or The crevice between the plies of a bolted weathering steel 5in. (127 mm). The limits on maximum pitch and edge distance joint usually seals itself with oxide if the joint is tight and does are ôomparable to those specified by AASHTO for all steel not move. If the joint moves, the surfaces should be coated with structures. These guidelines were confirmed by observations a protective material and filled with a suitable sealant to avoid made on 45 bolted joint specimens exposed about 7 years at progressive corrosions [Bethlehem 1983]. Kure Beach, North Carolina, and Monroeville, Pennsylvania The proper design of high-strength bolted weathering steel [Brockenbrough 1982]. Joint designs that met the suggested joints needs to consider: (1) corrosion of the contact surfaces; maximum pitch and edge distance developed little bowing be- and (2) slip resistance of friction-type joints with mill scale and tween bolts or edge distortion. balst cleaned surfaces. The first aspect is examined next; the To determine the expansive force of corrosion products, second aspect is examined in the sections under "Friction-Type Brockenbrough analyzed and tested tower joints removed from Joints with Mill Scale Surfaces" and "Friction-Type Joints with service. Each joint consisted of ,6 x 6 x /6 in main angles Blast-Cleaned Surfaces." spliced with 6 x 6 >< Y16 in splice angles 23/ in. (594 mm) Experience with bolted joints in exposed frameworks of long. Each leg was connected with four %-in. (16-mm) diameter

136

staggered bolts on either side of the splice center line on a 6- k= (58) in. (152-mm) pitch and 2/4-in. (57-mm) gage. Corrosion products 0.525 F, caused the corners of the angles to lift up by 0.43 in. (11 mm). Comparison of the measured lift-up with the displacement pre- The allowable shearing stress for friction-type joints is ob- dicted by a finite element analysis showed that the corrosion tained by substituting the allowable slip coefficient in Eq; 58 products developed an equivalent uniform pressure of 1.2 ksi and solving for: (8.3 MPa). At this pressure, the average measured bolt tensile stress was about 70 percent of the measured bolt tensile strength, F, = 0.525 F, k,,11 (59) reaching 95 percent for the bolt closest to the corner. Appar- ently, the bolts would have failed had corrosion continued. It The tensile strength for A325 bolts is F. = 120 ksi (830 MPa) should be kept in mind that these bolts were tightened snug and for Y, to 1 in. (12 to 25 mm) diameter, and F, = 105 ksi (725 widely spaced. MPa) for to l 2 in. (28 to 38 mm) diameter. For A490 bolts, Yura and Frank reported a 0.8-mil (20-rn) average increase lX it is F. = 150 ksi (1,035 MPa). in elongation of A325 Type 3 bolts in A588 steel joints after one year weathering [Yura 1981]. At that rate, and considering the ductility of the bolt, it would take about 100 years to fail the bolt in tension. They concluded from this observation that the expansive force caused by corrosion of the contact surfaces AASHTO Allowable Shear Stresses will not adversely affect the bolt performance in A588 steel joints with bolts placed in compact patterns. In 1957, AASHTO specified, for the first time, an allowable The initial bolt tension, the AASHTO limitations on bolt shear stress of 13.5 ksi (93 MPa) for A325 bolts. In 1965, they spacing, and the generally thicker sections should make bridge specified an allowable shear stress of 18 ksi (124 MPa) for A490 girder joints less susceptible to crevice corrosion than the tower bolts. These values, shown in Figure 102 with solid and dashed joints examined by Brockenbrough. lines, respectively, were based on early data for A7 and A440 Knotkova et al. reported that sufficiently tightened bolted steels joints on girder models exhibited no significant corrosive attack In 1974, Fisher and Struik surveyed the literature and re- even when exposed in highly aggressive environments [Knot- ported that for 180 tests of A7, A36, and A440 steel specimens kova 1982a]. The character of the external and internal joint with clean mill scale surfaces the average slip coefficient was surfaces, after pickling, was not significantly different. The rate = 0.322 and the standard deviation s = 0.062 [Fisher 1974]. of corrosion within the joint was not excessive. The quantity of They also reported that the mean clamping force measured in rust measured on the internal surfaces was smaller than that laboratory tests was higher than the minimum specified bolt on the external surfaces. The most suitable bolting material was tension, by 35 percent for turn-of-nut tightening, and by 13 found to be high-strength low-alloy weathering steel. Chrome- percent for calibrated wrench tightening. On the basis of the plated high-strength bolts were less suitable, and standard struc- Fisher [1974] findings, and assuming a 5 percent slip probability, tural steel bolts were least suitable. Galvanized bolts were also the allowable slip coefficient was calculated from reported to be unsuitable for weathering steel joints and are not recommended by the investigators for use in heavily polluted = 1.13 (Ic, = 1.645 s) (60) atmospheres. k,,11

Accordingly, AASHTO in 1979 increased the allowable shear stress for clean mill scale surfaces of all steels to F; = 0.525 x 120 x 1.13 (0.322 - 1.645 x 0.062) = 15.7 ksi = 16 ksi (110 FRICTION-TYPE JOINTS WITH MILL SCALE SURFACES MPa) for A325 bolts, and to F, = 20 ksi (138 MPa) for A490 bolts (Eq. 59). These values are plotted in Figure 102. Slip Resistance In 1981, Frank and Yura developed statistically reliable slip resistance data for bolted joints with coated contact surfaces. The slip resistance of high-strength bolted joints depends on In that study, they also proposed that the allowable shear stress the surface condition of the steel plates. The slip coefficient is for friction-type joints should be based on a 1 percent slip determined from experimental data using the expression: probability or a mean safety factor of 1.45, whichever gave the smaller value [Frank 1981], that is:

k,= — -- (57) nm Ti k,, = k, - 2.326 s (61)

in which: F, equals measured slip load, n equals number of bolts, or: m equals number of shearing planes, and T, equals initial bolt tension. The slip load can also be expressed in terms of a fic- k,,11 (62) titious slip shear stress on the gross area of all bolts, that is F = = n m .f, A,,. The specified minimum bolt tension is T, = 0.70 F, A,, in which A, is the tensile stress area, and F. is the tensile In their opinion, the possible increase of the clamping force strength of bolt. Substituting F and Ti in Eq. 57, and recalling above the minimum specified bolt tension should not be con- that A,/A,, 0.75, gives: sidered in the development of design recommendations. 137

NONWEATHERED MILL SCALE SURFACES

o Hechtmon 1955 Allowable Shear Stress V Bendigo 1963 -A325 Bolts A490 Bolts 8 ® Fisher 1963 - -- O Chesson 1965 1. Sterling 1966 O \sarhelyi 1967 A490 A325 o Allan 1968 Bolts Bolts D Birkemoe 1969 50 41 6 + Shoukry 1970 X Birkemoe 1971 o Frank 1981 Yuro 1981 40 30

0 4 30

lr-n 20 I - 20 2 t 10 10 I%lTYP) 0 REC0MMEND7 A588 STEE ID )Yura 1981) .4

D I - 1950 1960 1970 1980 1990

YEAR OF REPORT

Figure 102. Slip coefficients and allowable shear stresses for friction- type bolted joints with non weathered mill scale surfaces.

In a companion study, Yura et al. [Yura 1981] found that the 5 percent (1.645 s) and 1 percent (2.326 s) levels of slip the mill scale on A588 weathering steel was, on average, much probability. less resistant to slip ( = 0.23) than the mill scale on A7, A36 The second part of Table 64 summarizes the same data by and A440 steels (k, = 0.322) [Fisher 1974]. After analyzing type of steel. The allowable shear stresses were calculated for their data with Eqs. 59, 61, and 62, Yura, et al., recommended each series of tests and for various combinations of type of steel. that a new Class of Surface J for A588 clean mill scale be added The column for 5 percent slip probability, exluding the factor to Table 1.7.41C2 of the AASHTO specifications, with F = of 1.13 for increased bolt tension (Eq. 60), corresponds to the 10 ksi (69 MPa) for A325 bolts and F = 13 ksi (90 MPa) for AASHTO criterion. The columns for 1 percent slip probability A490 bolts. These proposed values are also plotted in Figure (Eq. 61) and 1.45 safety factor (Eq. 62) correspond to the criteria 102. proposed by Frank and Yura [Frank 1981]. The analysis of all previous data in terms of the new criteria suggests the following allowable shear stresses for clean mill scale surfaces. Review of Previous Data

ALLOWABLE SHEAR STRESS The decrease in allowable shear stress for A588 steel with F [ksi (MPa)] clean mill scale surfaces is due in large part to a reduced slip TYPE OF STEEL A325 BOLTS A490 BOLTS resistance and in small part to a change in the criteria for calculating the allowable shear stress. To place the values in Based on Frank and proper perspective, it was necessary to collect and analyze all Yura 's criteria: slip resistance data available to date. A36 carbon steel 13.5 (93) 17.0 (116) The first part of Table 64 summarizes the data for nonweath- A572 and A588 ered mill scale surfaces that were obtained from 254 tests of HSLA steels 10.0 (69) 12.5 (86) A514 quenched and A7, A36, A440, A572, A588, and A514 steel specimens. They tempered steel 9.0 (62) 11.3 (76) were obtained from 12 domestic studies. For each series of tests Based on Fisher and entered in the table, a data point was plotted in chronological Struik criteria: order in Figure 102. The vertical dimension line below each All steels 16.0 (110) 20.0 (138) point indicates the data scatter. The tick marks were drawn at 138

Table 64. Slip coefficients and allowable shear stresses for friction-type bolted joints with nonweathered mill scale surfaces.

Reference Type of Weathering No. of Average Slip Standard Allowable Shear Stress for A325 BoltsC Steel Time Tests Coefficient Deviation 5% 1% F05 = 1.45 (days) n is s (ksi) (ksi) (ksi)

Nonweathered Specimens

Hechtxian 1955 A7 0 58 0.342 0.046 16.8 14.8 14.8 Bendigo 1963 A7 0 20 0.359 0.101 12.1 7.8 15.6 Fisher 1963 A440 0 14 0.316 0.033 16.5 15.1 13.7 Chesson 1965 A7 0 11 0.262 0.043 12.0 10.2 11.4 Sterling 1966 8440 0 8 0.348 0.026 19.2 18.1 15.1 Vasarhelyl 1967 836, A440 U b 0.753 U.U33 12.5 11.1 11.0 A7 0 17 0.278 0.054 11.9 9.6 12.1 Ti 0 2 0.28 ------12.2 Allan 1968 836 0 8 0.285 0.014 16.5 15.9 12.4 Birkemoe 1969 8514 0 10 0.226 0.037 10.4 8.8 9.8 Shoukry 1970 826 0 39 0.32 0.044 15.6 13.7 13.9 Birke,noe 1971 8440 . 0 19 0.18 0.018 9.5 8.7 7.8 Frank 1981 836 0 6 0.22 ------15.2 8572 0 6 0.22 0.015 12.3 11.7 9.6 P514 0 2 0.30 13.0 0015d------Yura 1981 A588 0 208 0.250 14.2 13.5 0012d 10.9 0 ii 0.20 11.3 10.8 8.7 Ssinnary of Nonweathered Specimens

A7.A36,A440 0 203 0.306 0.071 11.9 8.9 13.3 836 0 52 0.313 0.043 15.3 13.4 13.6 8572 0 6 0.217 0.015 12.1 11.5 9.4 11588 0 31 0.23 0.034 11.0 9.5 10.0 A572,A588 0 37 0.23 0.015 12.9 12.3 10.0 A514 0 14 0.244 0.043 10.9 9.0 10.6

NOTES: a. Specimens tested at Federal Highway Administration. Specimens tested at University of Texas. For A490 bolts multiply shear stresses by 150 ksi/120 ksi = 1.25. Based on sanpie means and sample standard deviations [Natrella 1963].

The data for A7 and A44.0 steels were deleted because: (1) The data are summarized in Table 65 and plotted in Figure ASTM discontinued these specifications in 1968 and 1979, re- 103. The mean slip coefficient for A588 plates that were bolted spectively; (2) these steels have not been used in bridge con- and then weathered at most to one year did not appreciably struction for many years; and (3) data for obsolete steels should change with time of exposure. However, the mean slip coefficient not be allowed to influence the results for other steels when, as increased significantly when the plates were first weathered and it seems, slip resistance is to some degree a function of the mill then bolted prior to testing. The rate of increase was much scale that forms on different types of steel. That leaves A36, a greater for Fe 37 and Fe 52 steels than for A588 steel. Evidently, carbon steel, with the same allowable shear stress used prior to it takes more time for A588 to lose its hard mill scale through 1979. All others would have lower allowable stresses. corrosion. Both HSLA steels, A572 and A588, should have the same The results of the calculations summarized in Table 65 show allowable shear stress. For convenience of limiting the number that the allowable shear stresses for nonweathered mill scale of classes of surface, A514 steel could be lumped with the HSLA surfaces can also be applied to their weathered counterparts. steels. The allowable shear stresses for A709 Grade 100W weath- The small apparent losses exhibited by the bolted and then ering steel should be set at the same level as those for A5 14 weathered A588 steel specimens could reflect normal variations steel. between small samples rather than effect of weathering. There The aforementioned discussion showed that the lower mean appears to be no need to protect the mill scale contact surfaces slip resistance and a change in criteria have led to the reduced against corrosion, from the time when the steel members leave allowable shear stresses for A588 steel proposed by Yura et al. the shop until they are bolted at the bridge site. There is an obvious need to choose one set of criteria for calculating allowable shear stresses and to apply them uniformly to all steels, not just weathering steel. FRICTION-TYPE JOINTS WITH BLAST-CLEANED SURFACES

Weathered Mill Scale Surfaces AASHTO Allowable Shear,Stresses

Slip resistance data for steel plates with weathered mill scale Prior to 1979, the allowable shear stresses for friction-type surfaces are available from two studies [Kuperus 1966, Yura joints with blast-cleaned surfaces were the same as those for 1981]. The former was performed on European Fe 37 and Fe mill scale surfaces. These values ofF, = 13.5 ksi (93 MPa) for 50 steels with no enhanced atmospheric corrosion resistance, A325 bolts and F, = 18 ksi (124 MPa) for A490 bolts, shown the latter on A588 steel. The specimens were: (1) bolted and in Figure 104 with solid and dashed lines, respectively, were then weathered; or (2) weathered and then bolted. based on early data for A7 and A36 steels. 139

Table 65. Slip coefficients and allowable shear stresses for friction-type bolted joints with weathered mill scale surfaces.

Reference Type of Weathering No. of Average Slip Standard Allowable Shear Stress for A325 BoltsC Steel Time Tests Coefficient Deviation 5% 1% FOS = 1.45 (days) n k5 5 (ksi) (ksi) (ksi)

Bolted and Then Weathered Specimens 6b yurab 1981 A588 0 0.20 0.015 11.0 10.4 8.7 A588 180 5 0.18 0.004 0.91 10.7 7.8 A588 360 5 0.19 0.011 10.8 9.7 8.2 Weathered and Then Bolted Specimens

Kuperus 1966 Fe 37 0 4 0.35 0.055 16.3 14.0 15.2 Fe 37 5 4 0.37 0.040 19.1 17.4 16.1 Fe 37 14 4 0.48 0.020 28.1 27.3 20.8 Fe 37 60 4 0.55 0.020 32.5 31.7 23.9 Fe 52 0 4 0.25 0.020 13.7 12.8 10.9 Fe 52 5 4 0.27 0.010 16.0 15.5 11.7 Fe 52 14 4 0.35 0.030 18.9 17.6 15.2 Fe 52 60 4 0.51 0.06 25.9 23.3 22.1 5b Vura 1981 A588 0 0.20 0.009 11.7 11.3 8.7 A588 180 4 0.27 0.009 16.1 15.7 11.7 P588 360 5 0.37 0.068 16.2 13.3 16.1

NOTES: a. Specimen tested at Federal Highway Adminstratlon. Specimens tested at University of Texas. For A490 bolts multiply shear stresses by 150 ksi/120 ksi = 1.25.

I-O fl

WEATHERED MILL SCALE SURFACES

Kuperus l966Fe37 -.Kuperus 1966 Fe 32 Vuro 1981 A588

0-8: ---Bolted and Then Weathered -Weathered and Then Bolted

A490 A325 Bolts Bolts -150 -t40 0-6

40 3U

I- (0 20 w I 20 (C) A A S H TO All Steels

ID REC0MMENDED A 588 Steel (Yuro 981)

2 4 6 8 10 12

WEATHERING TIME, MONTHS

Figure 103. Slip coefficients and allowable shear stresses for friction- type bolted joints with weathered mill scale surfaces. 140

5, NONWEATHERED BLAST CLEANED SURFACES o Hechtmon 1955 8 • Douty 1965 O Chesson 1966 O Fisher 1968 o Frank 1981 Vuro 1981 A490 Bolts Albrecht 1983 Bolts 0 sol 40 0 6 1

j4oJ 30

U) U) Ui On ,30 .. U) i1 20 5%(TYP.) Ui U, TvP. I Quenched&J20 I 0. Tempered - All Steels Steels I -110 Allowable Shear Stresses: 10 - A325 Bolts -- A490 Bolts

1950 1960 1970 . 1980 1990

YEAR OF REPORT

Figure 104. Slip coefficients and allowable shear stresses for friction- type bolted joints with non weathered blast-cleaned surfaces.

In 1974, Fisher and Struik reported that the average slip poses of comparison. For each series of tests entered in the first coefficient of 168 grit-blasted Al, A36, and European Fe 37 part of the table, a data point was plotted in Figure 104 in steel specimens tested in several studies was k, = 0.493, with chronological order. The vertical dimensions below each point standard deviation s = 0.074 [Fisher 1974]. The corresponding indicate the data scatter, with tick marks drawn at the 5 percent values for 19 grit-blasted A5 14 steel specimens were k = 0.331 and 1 percent levels of slip probability. and s = 0.043. The second part of Table 66 summarizes the data by type of It appears that the allowable shear stresses in AASHTO's steel. The calculations of allowable shear stresses were based on 1979 Interim Specification were calculated from the aforemen- the criteria described in the previous section. On the basis of tioned data, assuming a 5 percent probability of slip and ne- the recommendations by Frank and Yura [Frank 1981], the data glecting the factor of 1.13 increase in bolt tension over the indicate the following allowable shear stresses for blast-cleaned minimum specified value. Substituting the means and standard surfaces: deviations in Eq. 60, one obtains from Eq. 59 for Class of Surface B blast-cleaned carbon and low-alloy steel and A325 bolts: ALLOWABLE SHEAR STRESSES, F = 0.525 x 120 (0.493 - 1.645 x 0.074) = 23.4 ksi F [ksi (MPa)] 25 ksi (172 MPa) TYPE OF STEEL A345 BOLTS A490 BOLTS and for Class of Surface C blast-cleaned quenched and tempered Based on Frank and steel and A325 bolts: Yura's criteria: A36 carbon steel 17.6 (121) 22.0 (152) F = 0.525 x 120 (0.331 - 1.645 x 0.043) = 16.4 ksi A572 and A588 HSLA steels 22.2 (153) 27.8 (191) 17 ksi (117 MPa) A514 quenched and tempered steel 14.4 (99) 18.0 (124) These values are also shown in Figure 104. Based on Fisher and Struik n criteria: Review of Previous Data Carbon and low- alloy steels 25.0 (172) 31.0 (214) All slip resistance data for nonweathered blast-cleaned sur- Quenched and faces were collected and are summarized in Table 66 for pur- tempered steels 17.0 (117) 21.0 (145) 141

Table 66. Slip coefficients and allowable shear stresses for friction-type bolted joints with nonweathered blast-cleaned surfaces.

Reference Type of Weathering No. of Average Slip Standard Allowable Shear Stress for A325 Bolts Steel Time Tests Coefficient Deviation 5% 1% FOS = 1.45 (days) n Rs s (ksi) (ksi) (ksi)

Nonweathered Specimens

Hechtman 1955 A7 0 3 0.47 0.108 18.4 13.8 20.4 Douty 1965 A7 0 7 0.560 0.068 28.2 25.3 24.3 Chesson 1966 P36 0 5 0.488 0.064 24.1 21.5 21.2 Kulak 1968 P514 0 17 0.332 0.041 16.7 14.9 14.4 Frank 1981 P36 0 21 0.495 0.097 21.1 17.0 21.5 P514 0 24 0.50 0.077 23.5 20.2 21.7 A572 0 68 0.561 0.097 25.3 21.1 24.4 5b Vura 1981 P588 0 0.86 0.038 50.2 48.6 37.3 20m 0 0.518 0.10 22.2 18.0 22.5 Albrecht 1983d A588 0 9 0.621 0.086 30.2 26.5 27.0

Summary of Nonweathered Specimens

A7 836 0 36 0.507 0.089 22.7 18.9 22.0 P36 0 26 0.492 0.091 21.5 17.6 21.4 A572 0 68 0.561 0.097 25.3 21.1 24.4 0090d A588 0 34 0.596 28.2 24.3 25.9 0095d A572, P588 0 102 0.573 26.3 22.2 24.9 P514 0 41 0.43 0.106 16.1 11.5 18.7

Notes: a. Specimens tested at Federal Highway Administration. b. Specimens tested at University of Texan. C. For A490 bolts multiply shear Stress by 150 ksi/120 ksi = 1.25. d. Based on Sample means and sample standard deviations [Natrella 1963].

The data for A7 steel were deleted for the reasons cited in imens had low slip resistance; and (2) a decrease when the the previous section. The HSLA steels have a higher slip re- control specimens had high slip resistance. sistance than carbon steels. The data for nonatmospheric corrosion resistant steels Fe 37, The data for A514 steels came from two series of tests with Fe 52, A36 and Grade 50 do not exhibit a consistent trend. vastly different results, namely: k, = 0.332 and s = 0.041 [Fisher Because bolted weathering steel joints of bridge members would 1968], versus k = 0.50 and s = 0.077 [Frank 1981]. The normally remain unassembled for, say, at most 3 months, cor- combination of the two sets would give an unrealistically high rosion of unprotected blast-cleaned contact surfaces should not standard deviation and low allowable shear stress. For this rea- affect the slip resistance. son, the allowable shear stresses for A5 14 steel in the previous The slip resistance of joints that were bolted and then weath- table were based on Fisher's data alone. These stresses should ered (dashed line in Figure 105) decreased slightly over a one- also apply to A709 Grade 100W steel. year weathering period. Longer exposures are needed to assess The need for uniform criteria for determining allowable shear the slip behavior of bolted joints in service. stresses, cited under "Friction-Type Joints with Mill Scale Sur- faces," is again evident. PIN AND HANGER CONNECTIONS

Weathered Blast-Cleaned Surfaces The crevice between two plies of a bolted weathering steel joint seals itself with corrosion products around the periphery Four studies examined the slip resistance of blast-cleaned Fe of the joint, if the bolt is tight and the joint stiffness is adequate 37, Fe 52, and A36 Grade 50 steels with no enhanced atmo- (see section under "High-Strength Bolted Joints"). However, spheric corrosion resistance; as well as A588 and British weath- progressive crevice corrosion occurs when the joints are loose. ering steels [Kuperus 1966, Lee 1969, Moss 1979, Yura 1981]. This was found in pin-and-hanger joints of bridge girders and Most specimens were weathered and then bolted; only one series in bolted highway guardrail joints with widely spaced bolts [Culp of specimens was bolted and then weathered. Weathering times 1980, McCrum 1980]. were at most one year. The space between the contact surfaces and between the pin The data are summarized in Table 67 and plotted in Figure and the hole of pin-and-hanger connections was reported to be 105. The trend for weathering steel specimens that were weath- tightly packed with dense rust, which could not be removed ered and then bolted indicates: (1) an increase in mean slip without disassembling the joint. This has occurred on bare resistance with weathering time when the 0-year control spec- weathering steel bridges after only 7 years of service. In some

Table 67. Slip coefficients and allowable shear stresses for friction-type bolted joints with weathered blast- cleaned surfaces.

Reference Type of Weathering No. of Average Slip Standard Allowable Shear Stress for P325 Bolts Steel Time Tests Coefficient Deviation 5% 1% FOS = 1.45 (days) n k s 5 (ksi) (ksi) (ksi)

Bolted and then Weathered Specimens

yurab 1981 A588 0 5 0.86 0.038 50.2 48.6 37.3 P588 180 5 0.77 0.035 44.8 43.3 33.4 P588 360 5 0.75 0.020 45.1 44.3 32.6

Weathered and then Bolted Specimens

Kuperus 1965 Fe 37 0 16 0.549 0.060 28.3 25.8 23.8 WEATHERED BLAST CLEANED SURFACES IC Fe 37 7 16 0.510 0.050 26.9 24.8 22.1 - Weathered and Then Bolted Fe 37 14 16 0.540 0.035 30.4 28.9 23.4 -----Bolted and Then Weathered Fe 52 0 16 0.649 0.077 32.9 27.7 28.2 Fe 52 7 16 0.522 0.068 25.8 22.9 22.7 Fe 52 14 16 0.600 0.052 32.4 30.2 26.0

Lee 1969 835 60 1 0.43 ------18.7 180 1 0.47 ------20.4 360 1 0.39 ------16.9 Moss 1979 Grade 50 A490 A325 0 3 0.422 0.031 23.4 22.0 18.3 Bolts Bolts Grade 50e 360 3 0.490 0.085 22.0 18.4 21.3 50 40 Grade 50f 360 3 0.545 0.016 32.6 32.0 23.7 Weathering 0 3 0.367 0.019 21.1 20.3 15.9 0.6V Weathering 360 2 0.473 ------20.5 1 Weathering 360 2 0.506 ------22.0 30 ' Yuraa 1981 P588 0 20 0062d 0.518 26.2 23.5 22.5 U) 8588 -90 0•066d U) 20 0.835 45.7 42.9 36.2 Si a:

U) fury6 1981 #588 0 5 0.86 0.038 50.2 48.6 37.3 20 a:< P588 180 5 0.74 0.043 42.1 40.3 32.1 Quenched e Tempered Steels A588 360 5 0.59 0.080 28.9 25.4 25.6 20 i) Kuperus 1966 Fe 37 Notes: a. Specimens tested at Federal Highway Administration. 02 - Kuperus 1966 Fe 52 Lee 1969 436 Specimens tested at University of Texas. Moss 1979 Grade 50 (Irdustrial Site) 10 For A490 bolts multiply shear stress by 150 ksi/120 Moss 1979 Grade 50 (Marine Site) ksi = 1.25. 0 10 O Moss 1979 Weathering Steel) Industrial Site) Based on sample means and sample standard deviations [Natrella 1963]. Moss 1979 Weathering Steel (Marine Site) Specimens weathered at industrial site. Yura 1981 4588 (Univ. Of Tenos) Specimens weathered at marine site. O Yura 1981 4588 (FHWA) 00 2 4 6 8 (0 12

WEATHERING TIME, MONTHS

Figure 105. Slip coefficient and allowable shear stresses for friction- type bolted joints with weathered blast-cleaned surfaces. 143

cases, corrosion was found to have frozen the joint, thus causing The spliced guardrail beams were tested in tension. Several other structural damage [Culp 1980]. modes of failure, and combinations thereof, were observed. The Pin-and-hanger connections should not be used in weathering most common failure occurred as the metal surrounding the steel bridges without adequate corrosion protection. Provisions lap joint bolts sheared and bent until one of the bolts could for cleaning, maintaining and painting such joints must be made. freely pass through the deformed holes of the lapped beams. Some of the joints failed in tension, usually along a line through HIGHWAY GUARDRAILS the bolt holes. The following conclusions were obtained from the test results: An examination of a weathering steel test stretch of guardrail after 15 years of exposure revealed that the lapped surfaces of The loss in strength of the galvanized guardrail after 15.5 the bolted joints had bulged out under the internal pressure of years of exposure at the Lansing site was negligible. The gal- the accumulated corrosion products [McCrum 1980]. To assess vanizéd coating remained intact. the corrosion-induced strength reduction of such joints, weath- The weathering steel guardrails lost 20 percent and 11 ering steel guardrail and galvanized guardrail sections were removed from a Lansing, Michigan, highway site after 15.5 years percent, respectively, of their initial strength when exposed 15.5 years at the Lansing site and 4.5 years at the Detroit site. of exposure, and from a Detroit, Michigan, urban site after 4.5 The joints that were removed intact had about the same years. Some joints were removed intact in order to evaluate the effect of the prestressed condition of the joint from tight packing strength as those that were disassembled and rebuilt. The expansive forces induced by the corrosion products did not seem- of corrosion products. Others were disassembled and rebuilt ingly affect joint strength. from end sections or interior sections of the guardrail beams in order to determine how differences in the two types of exposure (lapped versus freely exposed guardrail) affect joint strength. Guardrails can be built from weathering steel. But the contact Control joints were also built from new guardrail - beams for surfaces at the lap joints should be painted to protect them comparison with the joints in service. against crevice corrosion.

CHAPTER ELEVEN

STRUCTURAL DETAILS

BRIDGE DECK JOINTS number of such joints, by making multispan bridges continuous Flow of runoff water through open bridge deck joints or and sealing any deck joints that cannot be avoided. In recent leaking sealed joints has been one of the major causes of extensive years, cellular neoprene compression seals have been extensively maintenance and costly remedial work on bridges in general. used for sealed deck expansion joints of short-span bridges when The corrosion in these critical areas is particularly severe for the total movement does not exceed 2 to 2 2 in. (51 to 64 mm) weathering steel bridges that have no protective coating. and the skew angle is not extreme. For expansions greater than The desirable characteristics of a joint are water tightness, 2 in. (51 mm), designers have to rely on proprietary joint seals. smooth ridability, low noise level, wear resistance, and resistance Field experience indicates, however, that a large number of these to damaging snowplow blades. In reality, the performance of seals have performed poorly. The lack of water tightness has many joint systems is disappointing. When subjected to traffic brought the most complaints, although this feature is supposed and bridge movement, they fail in one or more important as- to be the most desired attribute of sealed joints [NCHRP 1979]. pects, notably water tightness. [NCHRP 1979]. To prevent progressive corrosion of weathering steel bridges, In some bridges, troughs were placed below open expansion several states have been painting the girder ends on either side joints to collect the runoff water and discharge it through drain over a length of one beam depth (Pennsylvania), 2 Y2 beam pipes away from the structure. This solution does not seem to depths (Illinois), or 5 to 10 ft (1.5 to 3.0 m) (Colorado, Louisiana, be viable because it introduces an additional item to clean and Maine, Maryland). Massachsuetts is also painting the cross maintain. The original problem reoccurs as soon as accumulated beams and stringer ends. road debris clogs the troughs and pipes and causes the runoff Weathering steel bridges must have water-tight joints. 0th- water to overflow. erwise the steel below the joint must be painted. To improve Conscious of the shortcomings of open expansion joints, many the rate of success in achieving water tightness, designers need designers have sought ways to reduce or even eliminate the to be knowledgeable about joint systems, inspectors must strictly 144

enforce construction quality control, and maintenance organi- ft (63-m) truss bridge is located away from industrial centers, zations should be responsible, capable, and adequately funded. in an area with 79 percent average relative humidity, 53-in. (1,340-mm) annual rainfall, and 52 F (11.3 C) average temperature. The designers have taken the following measures to ensure WATER AND DEBRIS ACCUMULATION that the protective oxide coating would form:

Water ponds and debris accumulate on horizontal surfaces Sloped the main trusses 5 percent out of plumb (Fig. 107). and in corners formed by horizontal and vertical plates. The Transversely sloped the lower flange of the floor beams most susceptible locations are: top of lower beam flanges and and the stringer flanges; and longitudinally sloped parts of the truss chords, gusset plates for horizontal bracing, bolted splices upper flange of the floor beams (Figs. 108 and 111). of horizontal and sloped members, at bearing and intermediate Selected box sections for all main truss and bracing truss stiffeners, and inside nonsealed box sections. To avoid water members (Figs. 108 to 110). ponding and debris accumulation it is important to: Sealed the box sections at both ends, and drilled drainage holes at the lowest point of the main truss diagonals (Fig. 112). Slope horizontal surfaces longitudinally or transversely. Extended the top flange of the upper chord beyond the Avoid reentrant corners. webs; extended the webs of both chords below the underside of Seal box sections and provide drainage holes at the lowest the bottom flange (Fig. 109). point. Mounted the rails on the stringers, without wooden railway Design details for easy discharge of water. ties (Fig. 111). Increased the thickness of the top flanges of stringers and Implementation of these measures is illustrated with the fol- floor beams by 80 mils (2 mm) as a corrosion allowance. The lowing examples of a truss bridge and a plate-girder bridge built Japanese National Railway specifications call for a 40-mils (1- in Japan. The information on these bridges was provided by mm) thickness increase when members are painted. Yamada [1983a]. Fabricated the conventional bearings from cast carbon steel and coated them with tar-epoxy-resin paint.

Railroad Truss Bridge Highway Plate Girder Bridge In 1980, the Japanese National Railways built their first nonpainted weathering steel railroad bridge, No. 3 Ohkawa Bridge, In 1980, the Hanshin-Express Highway Authority fabricated in Fukushima Prefecture, Northern Japan (Fig. 106). The 207- and assembled the Dejima Off-Ramp Bridge in Osaka, Japan. - 037 4.26 0.37

H.W.L. Span Length 63.0m

Figure 106. No. 3 Ohkawa Railroad Bridge.

Figure JOZ Cross-section of Ohkawa Railroad Truss Bridge.

145

ciU, 5° 2.5 O/e

A. Figure 108. Cross-section of top chord, stringer, and floor beam.

Stringer SECTION A Roor Beam'

0-0 Ln to -- _- 107 \

ure 109. Cross-section of top chord (left) and bottom 20 JL chordFig (right). _

5o,

Figure 110. Cross-section of top (left) and bottom (right) chord splices.

(a) (b)

Figure 111. Cross-section of stringer. Figure 112. Detail of bottom chord joint. 146

This three-span continuous weathering steel bridge is located in an industrial area. The following structural details are different from those used in painted bridges: Changed the cross sectional area of the lower flange by increasing thickness rather than width. Past experience had shown that uneven water flow at transitions in width caused nonuniform appearance and corrosion (Fig. 113). Increased the thickness of the lower flange plates by 40 mils (1 mm) as a corrosion allowance. Increased the gap between the flange ends at bolted splices to 0.8 in. (20 mm). Provided separate splice plates on the underside of the bottom flange for ease of drainage (Fig. 114). Eliminated reentrant corners at the bearing stiffeners by welding sloped plates, and provided sealed access holes (Fig. 115). Cut off the intermediate stiffeners 1.2 in. (30 mm) above the inside of the bottom flange, and coped the fitted stiffeners with 2-in. (50-mm) radius (Fig. 116). Figure 113. Water flow top-to-bottom surface at flange Bolted the tee-sections of the longitudinal bracing to the width transition. underside of the gusset plates. Enlarged the opening in the gusset plate through which the vertical stiffener passes (Fig. 117). Increased the height of the shoe base at the supports for better air circulation.

SECTION A

A Figure 114. Plan and elevation of bottom flange splice.

Acces

Figure 115. Closed-box detail at bearing stiffener. Figure 116 Intermediate stiffener detaiL 147

STAINING OF ABUTMENTS AND PIERS Opening

Weathering steel releases minute oxide particles of corrosion when water washes or drains over its surface. This oxide-laden rII runoff water stains the adjacent concrete abutments and piers, particularly during the early years of exposure. The rusty stains Y\ and streaks are unsightly. For aesthetic reasons every effort i, should be made to minimize them when the concrete supports SECTION A are visible to the public. < The best way to minimize staining is to incorporate permanent design details that will divert the runoff water away from the concrete. The following measures can be taken to prevent stain- Figure 117. Detail of longitudinal bracing joint. ing of concrete piers and abutments in new bridges:

Cover the pier caps with a polyethylene sheet until the bridge deck has been poured. The sheets can be removed once the deck is in place and a permanent system installed to carry away the rust-laden water. Build a parapet wall on top of piers and abutments. Drain the water through a pipe embedded in the concrete, or channel the overflow with a V-groove on vertical surfaces (Fig. 118) [Yamada 1983a]. Slope the abutment cap towards the retaining wall and Provide an 8-in. (200-mm) gap between the beams and the drain the water through a pipe into the dry well behind the wall ballast wall to imprOve ventilation and reduce the effects of (Fig. 119) [Bethlehem 1983]. In solutions (2) and (3) drain pipes moisture-induced deterioration on both steel and concrete must be maintained clean to avoid clogging. bridges. Install drip pans made from steel or fiberglass under the Do not use I-sections in grade separations because of the bearings and cantilever them out from the pier. When piers are possible accumulation of salt spray on the girders, especially in very high, wind can blow the dripping rust-laden water against areas of heavy salting. Box sections are permissible in grade the pier surface. separations. Attach a drip plate to the top of the lower flange and to Waterproof and pave concrete deck of box girders to pre- the girder web to divert wter off the steel before it drips on vent leakage through cracks in the deck slab. the concrete pier (Fig. 120). The Michigan experience shows Close diaphragms of box girders to keep out vermin and that drip plates do not prevent capillary, advancement of moisture. Ventilate box with a series of small holes in the bottom saltladen water [Arnold 1984]. flange. Insert a short tube in each hole, flush with the top surface Reduce the number of joints by making the bridge con- of the bottom flange, so that any seepage can escape without tinuous. running on the underside of the structure. Fit the tube with a 7: Coat concrete surfaces with liquid silicon sealer or other screen to prevent insects nesting in the box. formulations to prevent rust penetration into the concrete pores. Locate inspectior hatches in accessible places. Avoid lo- This method is not necessarily reliable. In some cases the treat- cations over the roadway or over ground which would not ment may discolor the concrete and oxide deposits may still support the inspection vehicle. For structures in grade separa- stain the concrete despite the treatment: Coatings break down tions, the preferred location is directly above the shoulder of with time. Two penetrant sealers examined in an experimental the highway. study did not adequately protect the concrete against oxide Extend the web of box girders below the bottom flange in stains and painted graffiti [Cosaboom 1978]. a manner similar to that shown in Figure 109. Given the potential for galvanic pitting corrosion along Stained surfaces of concrete piers and abutments of existing breaks in the mill scale and the very long time needed to form bridges can be either cleaned with chemical stain removers, a stable oxide film under sheltered conditions, blat clear the sandblasted, or painted dark brown. Allowing the stain to occur entire structure. freely for several years, and then treating the concrete in the Avoid ledges and horizontal surfaces wheremoisture and aforementioned ways, is not a satisfactory solution for new debris can accumulate. bridges because it adds one more maintenance task. Ventilate the superstructure well. Protect the weathering steel in the area of the expansion joints against corrosion either by controlling the flow path of RECOMMENDATIONS FROM ONTARIO STUDY runoff or by the use of a protective coating. Do not route drainage through closed sections. In a study conducted by the Ontario Ministry of Transpor- tation and Communications, the following recommendations Manning concluded that since remedial action is neither de- were made on how to improve design details so is to minimize sirable nor always effective, it is important to avoid those sites, the accelerated corrosion of weathering steel bridges [Manning section geometries, and design details that may result in poor 1984]: corrosion performance.

148

Figure 118. Sloping pier caps with a 4-in. (100-mm) high parapet wall. [Yamada 1983]

Girder

Drip Plate pPlate '

Figure 119. Sloping abutments and piers with drain. Figure 120. Drip plates attached to the bottom flange [Bethlehem 1983] and the web. [Bethlehem 1983] 149

CHAPTER TWELVE

CONCLUSIONS AND RECOMMENDED RESEARCH

CONCLUSIONS

The majority of the weathering steel bridges are in good condition, although localized areas of accelerated attack can be found in many structures. A number of bridges in several states are experiencing excessive corrosion because of salt contamination and/or prolonged time of wetness. Weathering steel can provide a satisfactory service life with limited maintenance if the structural details are designed in a manner that prevents accelerated attack, vulnerable areas are painted, and contamination with chlorides is avoided. Extensive information on the performance of weathering steel and its use in highway bridges has been assembled in this report. The practitioner can use this body of knowledge as a general guide for designing new bridges and for trouble shooting. The second phase of research under NCHRP Project 10-22 will develop specific guidelines for design, construction, maintenance, and rehabilitation of weathering steel bridges. The guide should include, as a minimum: (1) a quantitative defmition of corrosion resistance of weathering steels that is independent of the corrosion resistance of reference steels; (2) tolerable levels of corrosion penetration; (3) allowance for loss of section due to corrosion; (4) reduction in fatigue life due to pitting and corrosion fatigue; and (5) monitoring corrosion and pitting penetration of bridges in service.

RECOMMENDED RESEARCH

There are five major areas in which additional studies are addition to crack growth rate data, should be obtained for typical needed to properly assess the performance of nonpainted weath- details. In this way, the effect of corrosion on the total life ering steel and to establish the conditions under which the steel (initiation plus propagation) can be reliably determined in the can be used to its best advantage. These areas are corrosion, same test. fatigue, painting, joints, and specifications. The most important Evaluate joint seals for water tightness under movements topics that need to be studied are as follows: that typically occur in straight, skewed, and curved bridges. Determine the slip resistance and the effect of corrosion- Develop predictive models for estimating the corrosiveness induced expansive forces on the behavior of high-strength bolted of a site and the expected corrosion penetration of ideally ex- joints after long-term exposure. posed standard size coupon of weathering steel from known Develop effective methods of cleaning and painting cor- cliniatological and air pollution data. Determine how time-of roded weathering steel bridges, particularly those contaminated wetness can be correlated with climatological data. with salt. Assess the performance of the remedial painting sys- Relate the corrosion resistance measured under standard tem with natural exposure tests. Any such work should build conditions to the corrosion resistance of a large-size bridge girder on the results of the on-going studies being performed by the exposed under partially or fully sheltered conditions. Michigan DOT and the Structural Steel Painting Council. Determine the tolerable limits on time-of-wetness, air poi- lution, and deicing salt contamination for satisfactory performance of weathering steel. Perform corrosion fatigue tests of typical welded joints under conditions representative of bridge service environments. This should include large specimens, realistic cycling rates, variable amplitude loading, and intermittent spray. S-N data, in REFERENCES

AASHTO 1978 Guide Specifications for Fracture Critical Non-Redundant Albrecht 1982b Albrecht, P., "Fatigue Reliability Analysis of Highway Steel Bridge Members, American Association of State High- Bridges," Transportation Research Record 871, Transporta- way Officials, Washington, D.C., September 1978. tion Research Board, National Academy of Sciences, Washington, D.C., 1982.

AASHTO 1981 'Standard Specifications for Welding Structural Steel Highway Bridges, American Association of State Highway Albrecht 1983a Albrecht, P., and Cheng, J.G., "Fatigue of 8-Year and Transportation Officials, Washington, D.C., Third Weathered Automatically Welded A588 Steel Stiffeners," Edition, 1981. Journal of Structural Engineering, American Society of Civil Engineers, September 1983, pp. 2048-2065.

AASHTO 1983 'Weathering Steel Questionnaire,' Subcormiittee on Mainte- nance, American Association of State Highway and Trans- Albrecht 1983b Albrecht, P., "Fatigue Design Stresses for Weathering portation Officials, Washington, D.C., 1983. Steel Structures," Corrosion Fatigue: Mechanics, Metallurg', Electrochemistry, and En9ineering, ASTM STP 801, American Society for Testing and Materials, 1983, Abtahi 1976 Abtahi, A., Albrecht, P., and Irwin, G.R., 'Fatigue of pp. 445-471. Periodically Overloaded Stiffener Detail," Journal of the Structural Division, American Society of Civil Engineers, November 1976, pp. 2103-2119. Albrecht 1983c Albrecht, P., "Fatigue Design Stresses for Weathering Steel Structures," Closure of Discussion by J.W. Fisher, Corrosion Fatigue: Mechanics, Metallurgy, Electro- AFNOR 1979 "Structural Steels with Improved Atmospheric Corrosion chemistry and Engineering, ASTM STP 801, American Society Resistance--Sheets, Medium and Heavy Plates, Universal for Testing and Materials, 1983, pp. 445-471. Plates, Bars and Sections", AFNOR A35-502, French Standardization Association (AFNOR), Paris, France, 1979. Albrecht 1983d Albrecht, P., Sahli, A., Crute, D., Albrecht, Ph, and Evans, B., "Application of Adhesives to Bridge Structures,' Federal Highway Administration, Washington, Albrecht 1977 Albrecht, P., 'SEM Characterization of Naturally and D.C., 1984. - Artificially Weathered A588 Steel', Scanning Electron Microscopy, Vol. 1, lIT Research Institute, Chicago, Illinois, March 1977, pp. 719-724. Allan 1968 Allan, R.N., and Fisher, J.W., "Bolted Joints with Oversize or Slotted Holes,' Journal of the Structural Albrecht 1980 Albrecht, P., and Friedland, I.M., 'Fatigue Tests of Division, American Society of Civil Engineers, Vol. 94, 3-Year Weathered A588 Steel Weidment,' Journal of the ST9, September, 1968. Structural Division, American Society of Civil Engineers, Vol. 106, No. ST5, May 1980, pp. 991-1003. Allemeier 1981 Allemeier, K.A., "Unpainted Weathering Steel Bridges in Michigan," Paper presented at the Northwest Bridge Albrecht 1981 Albrecht, P., and Simon, S. "Fatigue Notch Factors for Engineers Seminar, Boise, Idano, October, 1981. Structural Details," Journal of the Structural Division, American Society of Civil Engineers, Vol. 107, No. ST7, July 1981, pp. 1279-1296. -Ang 1975 Ang, A. H-S., and Tang, W.H., "Probability Concepts in Engineering Planning and Design," Volume I - Basic Principles, John Wiley and Sons, 1975. Albrecht 1982a Albrecht, P., "Fatigue Behavior of Weathered Steel Bridge Components," Report FHWA/MD-81/02, Department of Civil Engineering, Universityof Maryland, College Park, Arnold 1980 Arnold, C.J., "Unpainted Weathering Steel Bridges in Maryland, Decerriber 1982. Michigan," Michigan Department of State Highways and Transportation, Testing and Research Division, Lansing, Michigan, 1980. Arnold 1981 Arnold, C.J., et al., "AISI Task Group on Weathering Baker 1982 Baker, E.A. and Lee, T.S., Calibration of Atmospheric Steel Committee Meeting of February 26, 1981, Michigan Corrosion Test Sites, ASTM STP 767, American Society for Department of State Highways and Transportation, Testing Testing and Materials, 1982, pp. 250-266. and Research Division, Lansing, Michigan, April 1981.

Baron 1983 Baron, J., 'Out of the Corrosion and into the Rust, Arnold 1984 Arnold, C.J., Private Conmiunication, Michigan Department Painting and Wallcovering Contractor, November 1983, of State Highways and Transportation, Testing and pp. 15-18. Research Division, Lansing, Michigan, 1984.

Barsom 1977 Barsom, J.M., and Novak, S.R., Subcritical Crack Growth ASTM 1968 'Corrosiveness of Various Atmospheric Test Sites as and Fracture of Bridge Steels," NCHRP Report 181, Trans- Measured by Specimens of Steel and Zinc," Metal portation Research Board-National Research Council, Corrosion in the Atmosphere, ASTM STP 435, American Washington, D.C., 1977. Society for Testing and Materials,1968, pp. 360-391.

Barsom 1983 Barsom, J.M., "Fatigue Behavior of Weathered Steel ASTM 1982 "Standard Recommended Practice for the Measurement of Components," U.S. Steel Corporation, Monroeville, Time-of-Wetness on Surfaces Exposed to Wetting Conditions Pennsylvania, January 1983. as in Atmospheric Corrosion Testing, ASTM Committee G-01.04, -April 1982 (Under Development). Bendigo 1963 Bendigo, R.A., Hansen, R.M. and Rumpf, ,].L., "Long Bolted Joints,' Journal of the Structural Division, American Atteridge 1982 Atteridge, D., Venkataraman, S., and Wood, W.E., Society of Civil Engineers, Vol. 89, ST6, December 1963. "Improving the Reliability and Integrity of Consumable Guide Electroslag Weldments in Bridge Structures", Federal Highway Administration, U.S. Department of Berger 1980 Berger, D.M., How Corrosion Theory Relates to Protective Transportation, December 1982. Coatings, Iron Steel Eng., Vol. 57, No. 8, August19a0, pp. 65-69.

AWS 1984 "Structural Welding Code - Steel, American Welding Society, Miami, Florida, 1984. Bethlehem Undated Bethlehem Steel Corporation, "Mayari R Weathering Steel", Booklet No. 2858, Bethlehem, Pennsylvania.

Bethlehem 1982 "Survey - Weathering Steel in Bridges," Bethlehem Steel Corporation, Bethlehem, Pennsylvania, December 1982.

Bethlehem 1983 "Mayari R Weathering Steel', Bethlehem Steel Corporation, Bethlehem, Pennsylvania, Booklet 3790, April 1983.

Birkemoe 1969 Birkemoe, P.C., Meinheit, D.E., and Munse, W.H., "Fatigue of A514 Steel in Bolted Connections," Journal of the Structural Division, American Society of Civil Engineers, Vol. 95, ST1O, October 1969.

Birkemoe 1971 Birkemoe, P.C., and Sninivason, R.S., Fatigue of Bolted High Strength Structural Steel," Journal of the Structural Division, American Society of Civil Engineers, Vol. 97, ST3, March 1971. Blake, G.T., Fatigue Tests of A588 Steel at Different Blake 1982 Brockenbrough 1982 Brockenbrough, R. L., and Gallagher, W. P., "Effect of Exposure Times During Six-Year Weathering." Unpublished tj Clamping Pressure and Joint Geometry on Bowing and Dis- Data, U.S.. Steel Corporation, Menroeville, Pennsylvania, tortion of Bare COR-TEN Steel Bolted Joints," Paper July 1982. presented at 11th Cor-Ten Conference, South Africa, June 1982. Boehni 1979 Boehni, V.11., Weathering Steel, Stahlbau/Korrosion, Schweizer Ingenieur and Architekt, No. 6, 1979 [in German] Brockenbrough 1983 Brockenbrough, R. L., "Considerations in the Design of Bolted Joints for Weathering Steel," Engineering Journal, Pmerican Institute of Steel Construction, First Quarter Bogar 1983 Bogar, F.D., and Crooker, T.W., 'Fatigue Testing in 1983, pp. 40-45. Natural Marine Environments and Substitute Ocean Waters," Materials Performanc"e, National Association of Corrosion Engineers, Vol. 22, No. 8, August 1983. Brooks 1983 Brooks, D., "Private Communication," Butler County, Ohio, 1983. Bohnenkamp 1973 Bohnenkamp, K., Burgmann, G. and Schwenk, W., "Investiga- tions of the Atmospheric Corrosion of Plain Carbon and Low Bukowiecki 1966 Schweizer Archivder Wissenschaftlichen Technik, Alloy Steels in Sea, Country and Industrial Air," Stahl Switzerland, Vol. 32, 1966 [in German]. und Eisen, Vol. 93, No. 22, October 1973, pp. 1054-1060 [in German]. Burgmann 1981 Burgmann, G., and Grimme, D., "Investigation of the Atmospheric Corrosion of Unalloyed and Low-Alloyed Steels Bonnarens 1981 Bonnarens, H., and Bragard, A., "Collective Research in Different Environments, Stahlbau, Vol. 50, No. 2, Prograli on Atmospheric Corrosion of Steels, Publication February 1981, pp. 56-59, [in German]. EUR 7400 Fr, Commission of the European Communities, Luxemburg, 1981 [in French]. Burnside 1983 Burnside, O.H., Hudak, J.J., Oelkers, E., and Chan, K., "Long-Term Corrosion Fatigue of Welded Marine Steels," Booth, G.S., "Corrosion Fatigue of Welded Steel Joints Booth 1983 Southwest Research Institute, San Antonio, Texas, May Under Narrow-Band Random Loading," in Corrosion Fatigue: 1983. Mechanics, Metallurgy, Electro-chemistry, and Engineering ASTM SIP 801, Mierican Society for Testing and Materials, 1983, pp. 472-490.

Boyd 1974 Boyd, W. K., et al., "Corrosion of Metals in the Atmo- sphere, Report No. MCIC-74-23, Metals and Ceramics Information Center, Battelle Columbus Laboratories, Columbus, Ohio, August 1974.

B.S. 1969 "British Standard Specification for Weldable Structural Steels, British Standard BS4360, 1969.

Bragard 1982 Bragard, A.A. and Bonnarens H.E., "Prediction at Long Terms of the Atmospheric Corrosion of Structural Steels from Short-Term Experimental Data, "ASTM SiP 767, Pmerican Society for Testing and Materials, 1982, pp. 339-358. Campbell 1970 Campbell, H.C., "Electroslag, Electrogas, and Related Copson 1959 Copson, H.R., and Larrabee, C.P., "Extra Durability of Welding Processes, Welding Research Council, Bulletin Paint on Low-Alloy Steels," ASTM Bulletin, Mierican No. 154, September 1970. Society for Testing and Materials, December 1959.

Carius 1930 Carius, C., Findings on Corrosion of Copper Steels, Copson 1960 Copson, H. R., "Long-Time Atmospheric Corrosion Tests on Zeitschrift fuer Metallkunde, Vol. 22, 1930, p. 337. Low-Alloy Steels, heerican Society for Testing and Materials," Vol. 60, 19609 pp. 650-667.

Chandler 1970 Chandler, K. A. and Kilcullen, M.B., 'Corrosion-Resistant Low-Alloy Steels: A Review with Particular Reference to Cosaboom 1978 Cosaboom, B., and Mehalchick, 0., 'Bridge Construction Atmospheric Conditions in the United Kingdom, British with Unpainted High Strength Low Alloy Steel: The Corrosion Journal, Vol. 5, No. 1, 1970, pp. 24-32. Aesthetic Problems,' New Jersey Department of Trans- portation, Division of Research and Development, Trenton, New Jersey, December 1978. Chesson 1965 Chesson, E., Jr., and Munse, W.H., "Studies of the Behavior of High-Strength Bolts and Bolted Joints," Engineering Experiment Bulletin 469, University of Cosaboom 1979 Cosaboom, B., Mehalchick, 0., and Zoccola, J.C., "Bridge Illinois, Urbana, 1965. Construction with Unpainted High-Strength Low-Alloy Steel: Eight-Year Progress Report," New Jersey Department of Transportation, Division of Research and Chesson 1966 Chesson, E., "Exploratory Tests of Structural Steel Development, Trenton, New Jersey, February 1979. Joints Assembled with Bolts and Adhesives," Civil Engineering Report, University of Illinois, Urbana, Illinois, 1966. Crosley 1983 Crosley, P., and Rippling, E., "Private Communication," Materials Research Laboratory, Glenwood, Illinois, 1983.

Circle 1983 Circle, J., 'Private Consiiunication, Franklin County, Col'inbus, Ohio, 1983. CSA 1976 'Structural Quality Steels, Standard 040.21, Canadian Standards Association, Rexdale, Ontario, Canada, June 1976. Coburn 1965 Coburn, S. K., "A Low-Cost Maintenance-Free Structural Steel for Highway Applications," Applied Research Laboratory, United States Steel Corporation, F'onroeville, CSSR 1978 Directions for Use of Atmofix Steels," Technical and Pennsylvania, 1965. Economic Research Institute of Metallurgical Industry, Prague, Czechoslovakia, December 1978.

Coburn 1974 Coburn, S. K., "Enhancing the Performance of Offshore Steel Structures," Offshore Technology Conference, Paper Culp 1976 Culp, J.D., 'Fracture Toughness and Fatigue Properties of No. OTC 2139. Texas, 1974, pp. 1081-1110. Steel Plate Butt Joints Welded by Submerged Arc and Electroslag Welding Procedures," Research Report No. R-1011, Michigan Department of State Highways and Coburn 1978 Coburn, S.K., "Increasing Container Service Life with Transportation, Lansing, Michigan, 1976. Painted USS COR-TEN Steel," Container Technology, Voliane II, Proceedings of 2nd Container Technology Conference, Brighton, England, December 1978, pp. 177-199. Culp 1980 Culp, J. D., and Tinklenberg, G.L., "Interim Report on Effect of Corrosion on Bridges of Unpainted A588 Steel and Paintp'l Steel Types, Michigan Department of State Copson 1945 Copson, H. R., 'A Theory of ne Mechanism of Rusting of Highways and Transportation, Testing and Research Low Alloy Steels in the Atmosphere," Pmerican Society for Division, Lansing, Michigan, June 1980. Testing and Materials, Vol. 45, 1945, pp. 579-579. DASt 1979 Supply, Fabrication, and Application of Weathering Edwards 1966 Edwards, A.M., Proc. Synp. Dev. Meth. Prev. Control 'Jl Steels," DASt Guideline 007, German Institute for Steel Corros. Bid., British Iron and Steel Federation, London, Construction, November 1979 [in German]. (1966).

DDR 1973 'Weather Resstant Structural Steels, Types of Steels, Ellis 1951 Ellis, 0. B., and LaQue, F.L., "Area Effects in Crevice General Technical Requirement," Standard TGL 28192, Corrosion, Corrosion, Vol. 7, No. 11, 1951, pp. German Democratic Republic,December 1973 [in German]. 362-364.

DDR 1983 "Guidelines for Application of Corrosion Resisting Steels Eurocode 1983 "Common Unified Code of Practice for Steel Structures, in Accordance with Standard TGL 28192," Private Section 9: Fatigue, ECCS/T61/M029, Eurocode 3, Draft, Comunication, Mvisory Office for Steel, Department of March 1983. Mining and Metallurgy, German Democratic Republic, 1983 [in German]. Evans 1972 Evans, U. R., et al., "Mechanism of Atmospheric Rusting,' Corrosion Science, Vol. 12, 1972, pp. 227-246. DFSL 1975 "Dofascoloy High Strength Low Alloy Steels," Dominion Foundries and Steel Limited, Hamilton, Ontario, Canada, 1975.

Douty 1965 Douty, RT., and McGuire, W.F., "High Strength Bolted Mement Connections," Journal of the Structural Division, Mierican Society of Civil Engineers, Vol. 91, ST2, April 1965.

DuMond 1977 Dut'bnd, T. C., "Sulfur Removal Gives Steel Superior Properties, Iron Age, Vol. 219, No. 17, May 1977.

Dunn 1983 Dunn, A. J., 'Private Communication,' Louisiana Department of Transportation and Development, Baton Rouge, Louisiana, 27 June 1983. Fisher 1963 Fisher, J.W., Ramseier, P., and Beedle, L.S., Strength Gallagher 1982 Gallagher, W.P., "Observations on the Condition of the of A440 Steel Joints Fastened with A325 Bolts," Weathering Steel Bridge, Louisiana Route 23 Over the Publications, IABSE, Vol. 23, 1963. Doullut Canal, Empire, Louisiana," Report to AISI Task Group on Weathering Steel Bridges, August 1982.

Fisher 1970 Fisher, J.W., Frank., K.H., Hirt, M.A., and McNamee, B.M., Effect of Weldments on the Fatigue Strength of Steel Garrido 1983 Garrido, L.A., "Private Communication," Louisiana Beams," NCHRP Report 102, Transportation Research Board- Department of Transportation and Development, Baton National Research Council, 1970. Rouge, Louisiana, 1983.

Fisher 1974 Fisher, J.W., Albrecht, P.A., Yen, B.T., Klingerman, Gurney 1980 Gurney, T.R., and Maddox, S.J., "Comparison of British D.J., and McNamee, B.M., Fatigue Strength of Steel Beams and Adierican Fatigue Design Rules for Welded Structures," with Welded Stiffeners and Attachments," NCHRP Report Preprint, Annual Convention and Exposition, Piiierican 147, Transportation Research Board-National Research Society of Civil Engineers, Holloywood-by-the Sea, Council, 1974. Florida, October 1980.

Fisher 1974 Fisher, J.W., and Struik, J.H.A., "Guide to Design Criteria for Bolted and Riveted Joints, John Wiley and Sons, 1974. -

Fisher 1983a Fisher, J.W., Private Communication, August 1983.

Fisher 1983b Fisher., J.W., "Fatigue Design Stresses for Weathering Steel Structures," Written Discussion of Paper by P. Albrecht, Corrosion Fatigue: Mechanics, Metallurgy, Electrochemistry, and Engineering, ASPI SIP 801, Pmerican Society for Testing and Materials, 1983, pp. 463-465.

Fletcher 1979 Fletcher, E. E., "A Review of the Status, Selection and Physical Metallurgy of High-Strength, Low-Alloy Steels," Report MCIC-79-39, Metals and Ceramics Information Center, Battelle Columbus Laboratories, Columbus, Ohio, March 1979.

Frank 1981 Frank,K.H. and Yura, J.A., "An Experimental Study of Bolted Shear Connections," Report FHWAJRD-81/148, Federal Highway Administration, U.S. Department of Transporta- tion, Washington, D.C. .205—December 1981.

Friedland 1982 Friedland, I.M., Albrecht, P.,•and Irwin, G.R., "Fatigue Behavior of 2-Year Weathered A588 Stiffeners and Attachments," Journal of the Structural Division, Mierican Society of Civil Engineers, Vol. 108, No. ST1, January 1982, pp. 125-144. Haynie 1978 Haynie, F., H., Spence, J.W. and Upham, J.B., Effects of Horton 1965 Horton, J.B., "The Rusting of Low Alloy Steels in the Air Pollutants on Weathering Steel and Galvanized Steel: Atmosphere," Presented at San Francisco Regional A Chamber Study," Atmospheric Factors Affecting the Technical Meeting, Anerican Iron and Steel Institute, Corrosion of Engineering Metals, ASTM SIP 646, Pmerican November 18, 1965. Society for Testing and Materials, 1978, pp. 30-47.

Hudson 1955 Hudson, J.C., and Stanners, J.F., Journal of Iron and Haynie 1982 Haynie, F. H., "Evaluation of the Effects of Microclimate Steel Institute, Vol. 180, No. 3, 1955. Differences on Corrosion," ASTM STP 767, Pinerican Society for Testing and Materials, 1982, pp. 286-308.

Hechtman 1955 Hechtman, R.A., Young, O.R., Chin, A.G., and Savikko, E.R., "Ship 06J01 Under Static Loads, ASCE Transactions, Vol. 120, 1955, pp. 1335-135.

Hein 1977 Hem, W., "Corrosion of Weathering Steels at Various Distances from the Sea," Report No. 42, Bundesanstalt fuer Wasserbau, Federal Republic of Germany, 1977 [in German].

Hein 1981 Rein, W. and Sczyslo, S., "Corrosion Behavior of Weathering Steels Under Deicing SaltExposure,', Interim Report No. 4, 1981.

Heroe 1975 Heroe, H., Evensen, J. and Emburg, J.D., "The Occurrence of Delamination in a Control Rolled HSLA Steel," Canadian Metallurgical Quarterly, Vol. 14; No. 2, April - June 1975 pp. 117-122.

Herzog 1960 Herzog, E., "Corrosion Resistant Steels," Corrosion- Anticorrosion, Vol. 8, No. 11, November 1960, pp. 394-406 [in French].

Hiaii 1978 Hiaii, J.R., and Pietrowski, R., The Influence of Forming and Corrosion on the Fatigue Behavior of Automotive Steels,' Society of Automotive Engineers, TPS No. 780040, March 1978.

Hiller 1966 Hiller, J.E., Steelwork and Corrosion, Vol. 17, 1966.

Hirt 371 Hirt, M.,'Fatigue Behavior of Rolled and Welded Beams, Ph.D. Dissertation, Lehigh University, Bethlehem, Pennsylvania, 1971.

Horikawa 1966 Horikawa, K., "Anti-Weathering Steel Plates Produced by Nippon Kokan Technical Research Institute, May 1966. JIS 1977 Rolled Steels for Welded Structure, 'Japanese Industrial Kasza 1983 Kasza, G.M., Private Communication, Federal. Highway Standard G3106, Japanese Standards Association, Tokyo, Administration, Region 10, Portland, Oregon, 1983. Japan, 1977.

Kilcullen 1979 Kilcullen, M.B., "Weathering Steels, Corrosion in Civil JIS 1981 "Hot Rolled Atmospheric Corrosion Resisting Steels for Engineering Proceedings, Institution of Civil Engineers, Welded Structure," G3114, JSSC, Vol. 17, No. 182, 1981. London, England, February 1979, pp. 95-105.

Jones 1937 Jones, J.A., Journal of Iron and Steel Institute, Vol. Knight 1977 Knight, J. W., "Corrosion Fatigue Related to Welded Steel 135, No. 1, 1937. Structure. A Literature Survey Welding Research Aboard Vol. 23, No. 9, November 1977. '

Knight 1983 Knight, R.P., "Economical Steel-Plate Girder Bridges," Slide Presentation, Bethlehem Steel Corporation, AISC National Engineering Conference, Memphis, Tennessee, March 1983.

Knotkova 1982a Knotkova, D., et. al., "Atmospheric Corrosion of Weathering Steels, ASTM STP 767, American Society for Testing and Material," pp. 7-44, 1982.

Knotkova 1982b Knotkova, 0., and Barton, K., "Corrosion Agressivity of Atmospheres (Derivation and Classification)," ASTM STP 767 American Society for Testing and Matrials, 1982, pp. 225-249.

Konkol 1979 Konkol, P.J. and Domis, W.F., "Causes of Grain-Boundary Separations in Electroslag Weld Metals," Welding Research Supplement, Welding Journal, June 1979, pp. 161-S to 167-S.

Kulak 1968 Kulak, G. and Fisher, J.W., 'A514 Steel Joints Fastened by A490 Bolts," Journal of the Structural Division, American Society of Civil Engineers, Vol. 94, ST10, October 1968, pp. 2303-2323.

Kunihiro 1972 Kunihiro, T., move, K., and Fukuda, T., "Atmospheric Exposure Study of Weathering Steel," Research Lab Report Br. 71-08, Ministry of Construction, Tokyo, Japan, 1972 [in Japanese].

Kunze 1974 Kunze, E. 'Electrochemical Studies of Rusting in Plain Carbon and Weathering Steels," Neue Huette, Vol. 19, No. - - 5, May 1974, pp. 295-303 [in German]. Kuperus 1966 Kuperus, A., The Ratio Between the Slip Factor of Fe 52 La Que, F.L., "Corrosion Testing," American Society for La Que 1951 LA and Fe 37, CEACM, X-6-27, Stevin Laboratory. Deift Testing and Materials, Proceedings, Vol. 51, 1951, pp. 00 University of Technology, The Netherlands, 1966. 495-582.

Larrabee 1941 Larrabee, C.P., The Effect of Specimen Position on Atmospheric Corrosion Testing of Steel, Transactions of Electrochemical Society, Vol. 85, 1941, pp. 297-303.

Larrabee 1958 Larrabee, C. P., "Corrosion-Resistant Experimental Steels for Marine Applications," Corrosion, Vol. 14, No. 11, 1958, pp. 501T, 1958.

Larrabee 1961 Larrabee,C.P., and Coburn, S.K., "The Atmospheric Corrosion of Steels as Influenced by Changes in Chemical Composition,' 1st International Congress on Metallic Corrosion, London, 1961, pp 276-285.

Larrabee 1966 Larrabee, C. P., 'Corrosion Resistance of High-Strength Low Alloy Steels as Influenced by Composition and Environment", U.S. Steel Corporation, Research and Development Laboratory, 1966.

Lee 1969 Lee, J.H., O'Connor, C., and Fisher, J.W., "Effect of Surface Coatings and Exposure on Slip," Journal of the Structural Division, hi,erican Society of Civil Engineers, Vol. 95, ST1, November 1969.

Legault 1974 Legault, R.A., and Leckie, H.P., 'Effect of Alloy Composition on the Atmospheric Corrosion Behavior of Steels Based on a Statistica Analysis of the, Larrabee-Coburn Data Set," Corrosion in Natural Environments, ASTM STP 558, Anerican Society for Testing and Materials, 1974, pp. 334-347. Manning 1984 Manning, D.G., Accelerated Corrosion in Weathering Steel Nakayama 1972 Nakayama, Y., "Durability of Paint on Atmospheric Bridges," Canadian Structural Engineering Conference, Corrosion Resistant Steels, Chemical Abstracts, Vol. 76, Canadian Steel Construction Council, Willowdale, OntariO 1972. 1984.

Natrella 1963 Natrella, M.G., "Experimental Statistics, Handbook No. McCrum 1980 r4cCrum, R. L., and Arnold, C.J., Weathering Steel 91, National Bureau of Standards, Washington, D.C., August Guardrail: A Materials Performance Evaluation," Michigan 1963. Department of Transportation, Testing and Research Division, Lansing, Michigan, November 1980. NCHRP 1979 'Bridge Deck Joint-Sealing Systems, Evaluation and Performance Specification, NCHRP Report 204, Transporta- McCri.mi 1983 tCrtiii, R.L., Unpainted A588 Corrosion Survey, Current tion Research BOard, National Research Council, Status," Office Memorandum, Michigan Department of Washington, D.C., June 1979. Transportation, Testing and Research Division, Lansing, Michigan, August 1983. Nihei 1978 Nihei, M., Yokda, M., and Sasaki, E., "Fatigue Properties for Butt Welded Joint of SM 50 A High Tensile Strength McKenzie 1978 ItKenzie, M., "The Corrosion Performance of Plate,' Transactions of National Research Institute for Weathering Steel in Highway Bridges," Report No. 857, Metals, Japan, Vol. 20, No. 4, 1978. Transport and Road Research Laboratory, Crowthorne, Berkshire, United Kingdom, 1978. Nippon 1974 "Characteristics of Cor-Ten Steel With Weather-Coat Coating," Nippon Steel Corporation, 8th International Misawa 1971 Misawa, T., et al., "The Mechanism of Atmospheric Rusting Cor-Ten Conference, Borlange, Sweden, June 1974. and the Effect of Cu and P on the Rust Formation of Low Alloy Steels," Corrosion Science, Vol. 11, No. 1, 1971, pp. 35-48. Nippon 1976 "Studies on Rust of Cor-Ten Steel with Weather-Coating," Nippon Steel Corporation, 9th International Cor-Ten Conference, Rio de janeiro, Brazil, October 1976. Misawa 1973 Misawa, T., "The Thermodynamic Consideration for the Iron-Water System at 25'C," Corrosion Science, Vol. 13, No. 9, September 1973, pp. 659-676. Noel 1972 Noel, J.S., and Toprac, A.A., "Static, Fatigue, and Impact Strength of Electroslag Weldments," Center for Highway Research, - University of Texas Research Report No. 157-1F, Misawa 1974a Misawa, T., et al., "The Mechanism of Formation of Iron December 1972. Oxide and Oxyhydroxides in Aqueous Solutions at Room Temperature," Corrosion Science, Vol. 14, 1974, pp. 131- 149. Norcross 1965 Norcross, J.E., "Electroslag/Electro-gas Welding," Welding Journal, Vol. 44, No. 3, March 1965, pp. 176-186. Misawa 1974b Misawa, T., et al., "The Mechanism of Atmospheric Rusting and Protective Miorphous Rust on Low Allow Steel," Novak 1983a Novak, S.R., "Private Communication," U.S. Steel Corpora- Corrosion Science, No. 14, 1974, pp. 279-89. tion, Menroeville, Pennsylvania, 1983.

Mess 1979b Moss, D.S., "High Strength Friction Grip Bolted Joints - Novak 1983b Novak, S.R., 'Corrosion-Fatigue Crack Initiation Behavior Effects under one Year of Weathering Under Load, Report of Four Structural Steels, Corrosion Fatigue: Mechanics, 499, Transport and Road Research Laboratory, Crowthorne, Metallurgy, Electrochemistry and Engineering ASTM STP 801, Berkshire, United Kingdom, 1979. American Society for Testing and Materials, 1983, pp. 26-63. Muta 1965 Muta, T. and Nishi, T., I.I.W. Document No. IX-448-65, Yawata Iron and Stel Co. Ltd., 1965. Otake 1962 Otaka, T., et al., Yawata Technical Report No. 238, March Paton 1962a Paton, B.E., "Electroslag Welding," Ptiierican Welding 1962. Society, 2nd Edition, New York, 1962. 0

Paton 1962b Paton, B.E., "Electroslag Welding of Very Thick Materials," Welding Journal, Vol. 41, No. 12, December 1962, pp. 1115-1122.

Petch 1959 Petch, N. J., 'The Ductile Cleavage Transmission in Alpha Iron," Fracture (B.C. Averbach, et al., Editors), Cam- bridge Technology Press of MIT, John Wiley and Sons, New York, N.Y., 1959, pp. 54-67.

Pickering 1963 Pickering, F. B., and Gladman, 1., 'An Investigation into Some Factors Which Control the Strength of Carbon Steels," Metallurgical Developments in Carbon Steels, ISI Special Report 81, The Iron and Steel Institute, London, England, 1963, pp. 10-20.

Pickering 1976 Pickering, F. B., "High-Strength Low-Alloy Steels - A Decade of Progress," Microalloying 75 Proceedings, 1976, pp. 3-24.

Pilling 1948 Pilling, N. B., and Wesley, W. A., Proceedings Pmerican Society for Testing and Materials, Vol. 48, 1948, pp. 610-617.

Porter 1982 Porter, L. F., and Repas, P.E.., "The Evolution of HSLA Steels," Journal of Metals, April 1982, pp. 14-21. Raska 1983 Raska, J.G., Private Communication," Materials and Test Scantlebury 1980 Scantlebury, J. D., Callow, L., and De Sousa, J. D., Division, State Department of Highways and Public "Inhibition of the Corrosion of Mild and Low-Alloy Steel Transportation, Austin, Texas, 1983. by Zinc Phosphate Pigments," Corrosion and Protection Center, University of Manchester Institute of Science and Technology, Manchester, England, 1980. Reed 1982 Reed, F. 0., and Kendrick, C.B, "Evaluation of Weathering Effects on Structural Steel," State of California Department of Transportation, Division of Construction, SCC 1973 "Structural Steel: Selection and Uses," 3rd Edition, The Office of Transportation Laboratory, Sacramento, Steel Company of Canada, Toronto, Ontario, July 1973. California, March 1982.

Schikorr 1963 Schikorr, G., Werkstoffe Und Korrosion Vol. 14, 1963, Roberts 1974 Roberts, R., Irwin, G.R., Krishna, G.V., and Yen, B.T., p. 69 [in German]. "Fracture Toughness of Bridge Steels, Report No. FHWA-RD-74-57, Federal Highway Adminstration, Washington, D.C., September 1974. Schikorr 1964 Schikorr, G., Werkstoffe Und Korrosion, Vol. 15, 1964, p. 457 [in German].

Roberts 1977 Roberts, R., et.al , "Determination of Tolerable Flaw Sizes in Full Size Welded Bridge Details," Report No. Schikorr 1967 Schikorr, G., Werkstoffe Und Korrosion, Vol. 18, 1967, FHWA-RD-77-170, Federal Highway Administration, p. 514, [in German]. Washington D.C., December 1977.

Schmitt 1967 Schmitt, R. J., and Mathay, W. L., "Tests Show Perform- Roberts 1983 Roberts, R., and Fisher, J.W., "Corrosion Fatigue ance of Low Alloy Steels in Chemical Plant Environments, Characteristics of Bridge Steels," Research in Progress Materials Protection, Vol. 6, No. 9, 1967, pp. 37-42. Sponsored by Federal Highway Administration, Washington, D.C., 1983. Schmitt 1969 Schmitt, R. J., and Phelps, E. H., "Corrosion Performance of Constructional Steels in Marine Applications," First Offshore Technology Conference, Houston, Texas, May 1969.

Schramm 1934a Schramm, G. N., Taylerson, E. S., and Steubing, A. F., - Iron Age, Vol. 134, 1934.

Schramm 1934b Schramm, G. N., and Taylerson, E. S., Syiiposium on Out- door Weathering of Metals and Metallic Coatings, Anerican Society for Testing and Materials, Philadelphia, Pennsylvania, 1934.

Schramm 1936 Schramm, G. N., Taylerson, E. S., and Larrabee, C. P., Railway Age, Vol. 101, 1936.

Schwartz 1972 Schwartz, H., Stelwork and Corrosion, Vol. 23, 1972. Scott 1975 Scott, P.M., and Silverster, D.R.V., "The Influence of Takeshi 1975 Takeshi, Y., "Larnellar Tearing and Marine Structures," Sea Water on Fatigue Crack Propagation Rates in Welding and Metal Fabrication, Vol. 43, No. 10, Structural Steel," Interim Technical Report UKOSRP 3/03, December 1975, pp. 740-746. Department of Energy, UK Offshore Steels Research Project, December 1975. Taylerson 1941 Taylerson, E.S., Journal of Iron and Steel Institute, Vol. 143, 1941. Sereda 1960 Sereda, P.J., "Measurement of Surface Woisture and Sulfur Dioxide Activity at Corrosion Sites," ASTM Bulletin, Pmerican Society for Testing and Materials, No. 246, May Taylor 1981 Taylor, M.E., and Barsom J.M., "Effect of Cyclic 1960, pp. 47-48. Frequency on the Corrosion-Fatigue Crack-Initiation Behavior of ASTM A517 Grade F Steel," Fracture Mechanics: Thirteenth Conference, ASTM STP 743, Pmerican Society for Sereda 1982 Sereda, P. J., et al., Measurement of the Time-of- Testing and Materials, 1981, pp. 599-622. Wetness by Meisture Sensors and Their Calibration," STP 767, I¼nerican Society for Testing and Materials, 1982, pp. 267-285. Tinklenberg 1982 Tinklenberg, G.L., 'Corrosion of Unpainted Weathering Steel," Second World Congress on Coating Systems for Bridges and Steel Structures, New York, New Serizawa 1973 Serizawa, S., et al., "Corroson Fatigue Prperties of York, October 1982. the Welded Joints in 50 kg/mm and 80 kg/!ml Class High Tensile Strength Steels,' Nippon Steel Corporation, Tokyo, Japan, September 1973. Tomkins 1982 Tomkins, B., "Fatigue Design Rules for Steel Welded Joints in Offshore Structures,', Paper No. OTC 4403, Offshore Technology Conference, Houston, Texas, May Shaaban 1983 Shaaban, H. and Albrecht, P., "Collection and Analysis of 1982. Stress Range Histograms Recorded on Highway Bridges," Civil Engineering Report, University of Maryland, College Park, Maryland, March 1983. Townsend 1982 Townsend H.E. and Zoccola, J.C., "Eight-Year Atmospheric Corrosion Performance of Weathering Steel in Industrial, Rural, and Marine Environments," Atmospheric Corrosion of Simon 1981 Simon, S., and Albrecht, P., "Adding Fatigue Life to Metals, ASTM STP 767, Mierican Society for Testing and Cover Plate Ends," Journal of the Structural Division, Materials, 1982, pp. 45-59. Mierican Soceity of Civil Engineers, Vol. 107, No. ST5, May 1981, pp. 923-935.

Smith 1983 Smith, R., "Private Communication," Tuscarawas County, New Philadelphia, Ohio, 1983.

Songa 1975 Songa, I., Proceedings, Information Meeting, European Steel and Coal Corruiunity, Luxemburg, 1975.

Sterling 1966 Sterling, G.H., and Fisher, J.W., "A440 Steel Joints Connnected by A490 Bolts," Journal of the Structural Division, Pmerican Society of Civil Engineers, Vol. 92, No. ST3, June 1966, p 101-118.

Suzuki 1980 Suzuki, I., Hisamatsu, Y., and Masuko, N., "Nature of Atmospheric Rust on Iron," Journal of Electrochemical Society, Vol. 127, No. 10, October 1980, pp. 2210-2215. Uhlig 1973 Uhlig, H.H., Distinguishing Characteristics of Pitting Van Eijnsbergen 1979 Van Eijnsbergen, J.F.H., "Weathering Steels Versus and Crevice Corrosion, Materials Protection and Ordinary Structural Steels as a Base for Paints, Cargo Performance, Vol. 12, No. 2, 1973, pp. 42-44. Systems International, Vol. 5, No. 6., 1979, pp. 101-103.

UK DOE 1981 Offshore Installations: Guidance on Design and Vannerberg 1976 Vannerberg, N. G., "The Effect of Small Alloying Construction,' Proposed New Fatigue Design Rules for Additions on the Oxidation of Steel," Metalloberflaeche, Steel Welded Joints in Offshore Structures, U.K. Vol. 30, No. 1, January 1976, pp. 18-20 [in German]. Department of Energy, 1981.

Vasarhelyi 1967 Vasarhelyi, D., and Chiang, K.C., 'Coefficient of UK DOE 1978 'Offshore Installations: Guidance on Design and Friction in Joints of Various Steels,' Journal of the Construction," U.K. Department of Energy, 1978. Structural Division, Pnierican Societyof Civil Engineers, Vol. 93, ST4, August 1967.

USS 1975a 'Short Span Steel Bridges,' Report ADUSS 88-4551-02, United States Steel, Pittsburgh, Pennsylvania, November Vernon 1931 Vernon, W.H.J., "A Laboratory Study of the Atmospheric 1975. Corrosion of Metals, Part I,' Transactions of Faraday Society, Vol. 27, March1931, pp. 255-277.

Vernon 1933 Vernon, W.H.J., 'The Role of the Corrosion Product in the Atmospheric Corrosion of Iron," Transactions of Electro- chemical Society, Vol. 64, August 1933, pp. 31-33.

Vernon 1935 Vernon, W.H.J., 'A Laboratory Study of the Atmospheric Corrosion of Metals, Part II,' Transactions of Faraday Society, Vol. 31, October 1935, pp. 1668-1701.

Vernon 1960 Vernon, W.H.J., Journal of Iron Steel Institute, Vol. 196, 1960, p. 333. Wilson 1982 Wilson, A.D., "Influence of Inclusions on the Fracture Yamada 1977 Yamada, K. and Albrecht, P., "Fatigue Behavior of Two Properties of A588-A Steel," 15th National Synposiuii on Flange Details," Journal of ;he Structural Division, Fracture Mechanics, University of Maryland, College Park, Mierican Society of Civil Engineers, Vol. 103, No. ST4, Maryland, July 1982. April 1977, pp. 781-791.

Woodley 1966 Woodley, C.C., et al., "Electroslag Welded Wide Plate Yamada 1983a Yamada, K., "Japanese Experience on Weathering Steel Tests on 3 in. Thick Mild Steel," British Welding Bridges," Department of Civil Engineering, Nagoya Journal, March 1966, pp. 165-173. University, 1983.

Yamada 1983b Yamada, K. and Kikuchi, V., "Fatigue Behavior of 2-Year and 4-Year Weathered Welded Joints,' Department of Civil Engineering, Nagoya University, Nagoya, Japan, February 1983.

Yazdani 1983 Yazdani, N. and Albrecht, P., "Crack Growth Rates of Structural Steels in Air and Aequous Environments," Civil Engineering Report, University of Maryland, College Park, Maryland, November 1983.

Yura 1981 Yura, J. A., et al., "Bolted Connections with Weathering Steel,' Journal of the Structural Division, Proc. Paper 16644, ASCE, Vol. 107, No. ST11, November, 1981, pp. 2071-2087.

Zoccola 1976 Zoccola, J.C., 'Eight Year Corrosion Test Report - Eight Mile Road Interchange,' Bethlehem Steel Corporation, Bethlehem, Pennsylvania, 1976. THE TRANSPORTATION RESEARCH BOARD is a unit of the National Research Council, which serves the National Academy of Sciences and the National Academy of En- gineering. The Board's purpose is to stimulate research concerning the nature and performance of transportation systems, to disseminate information that the research produces, and to en- courage the application of appropriate research findings. The Board's program is carried out by more than 270 committees, task forces, and panels composed of more than 3,300 administrators, engineers, social scientists, attorneys, educators, and others concerned with transportation; they serve without compensation. The program is supported by state transportation and highway departments, the modal administrations of the U.S. Depat*ment of Transportation, the Association of American Railroads, the National Highway Traffic Safety Administration, and other organizations and individuals interested in the development of transportation. The National Research Council was established by the National Academy of Sciences in 1916 to associate the broad community of science and technology with the Academy's purposes of furthering knowledge and of advising the Federal Government. The Research Council has become the principal operating agency of both the National Academy of Sciences and the National Academy of Engineering in the conduct of their services to the government, the public, and the scientific and engineering communities. It is administered jointly by both Academies and the Institute of Medicine. The National Academy of Sciences was established in 1863 by Act of Congress as a private, nonprofit, self-governing membership corporation for the furtherance of science and technology, required to advise the Federal Government upon request within its fields of competence. Under its corporate charter the Academy established the National Research Council in 1916, the National Academy of Engineering in 1964, and the Institute of Medicine in 1970. TRANSPORTATION RESEARCH BOARD National Research Council NON-PROFIT ORG. U.S. POSTAGE 2101 Constitution Avenue, N.W. PAID Washington, D.C. 20418 WASHINGTON, D.C.

ADDRESS CORRECTION REQUESTED PERMIT NO. 8970

fl- 0 co U- - Q

0 I- LU

LU (/)J m

0 oaw -0uJ V) 0l_- <0 00